A Self-Supervised Few-Shot Semantic Segmentation Method Based on Multi-Task Learning and Dense Attention Computation

Abstract

Nowadays, autonomous driving technology has become widely prevalent. The intelligent vehicles have been equipped with various sensors (e.g., vision sensors, LiDAR, depth cameras etc.). Among them, the vision systems with tailored semantic segmentation and perception algorithms play critical roles in scene understanding. However, the traditional supervised semantic segmentation needs a large number of pixel-level manual annotations to complete model training. Although few-shot methods reduce the annotation work to some extent, they are still labor intensive. In this paper, a self-supervised few-shot semantic segmentation method based on Multi-task Learning and Dense Attention Computation (dubbed MLDAC) is proposed. The salient part of an image is split into two parts; one of them serves as the support mask for few-shot segmentation, while cross-entropy losses are calculated between the other part and the entire region with the predicted results separately as multi-task learning so as to improve the model’s generalization ability. Swin Transformer is used as our backbone to extract feature maps at different scales. These feature maps are then input to multiple levels of dense attention computation blocks to enhance pixel-level correspondence. The final prediction results are obtained through inter-scale mixing and feature skip connection. The experimental results indicate that MLDAC obtains 55.1% and 26.8% one-shot mIoU self-supervised few-shot segmentation on the PASCAL- 5 i and COCO- 20 i datasets, respectively. In addition, it achieves 78.1% on the FSS-1000 few-shot dataset, proving its efficacy.

Full text

Citation: Yi, K.; Wang, W.; Zhang, Y.
A Self-Supervised Few-Shot Semantic
Segmentation Method Based on
Multi-Task Learning and Dense
Attention Computation. Sensors 2024,
24, 4975. https://doi.org/10.3390/
s24154975
Academic Editor: Marco Leo
Received: 31 May 2024
Revised: 21 June 2024
Accepted: 22 July 2024
Published: 31 July 2024
Correction Statement: This article
has been republished with a minor
change. The change does not affect
the scientific content of the article and
further details are available within the
backmatter of the website version of
this article.
Copyright: © 2024 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
sensors
Article
A Self-Supervised Few-Shot Semantic Segmentation Method
Based on Multi-Task Learning and Dense Attention Computation
Kai Yi 1, Weihang Wang 2and Yi Zhang 2,*
1Intelligent Policing Key Laboratory of Sichuan Province, Luzhou 646099, China; yikai@scpolicec.edu.cn
2College of Computer Science, Sichuan University, Chengdu 610042, China; wei_sailing@163.com
*Correspondence: yi.zhang@scu.edu.cn
Abstract: Nowadays, autonomous driving technology has become widely prevalent. The intelligent
vehicles have been equipped with various sensors (e.g. vision sensors, LiDAR, depth cameras etc.).
Among them, the vision systems with tailored semantic segmentation and perception algorithms play
critical roles in scene understanding. However, the traditional supervised semantic segmentation
needs a large number of pixel-level manual annotations to complete model training. Although few-
shot methods reduce the annotation work to some extent, they are still labor intensive. In this paper,
a self-supervised few-shot semantic segmentation method based on Multi-task Learning and Dense
Attention Computation (dubbed MLDAC) is proposed. The salient part of an image is split into two
parts; one of them serves as the support mask for few-shot segmentation, while cross-entropy losses
are calculated between the other part and the entire region with the predicted results separately as
multi-task learning so as to improve the model’s generalization ability. Swin Transformer is used
as our backbone to extract feature maps at different scales. These feature maps are then input to
multiple levels of dense attention computation blocks to enhance pixel-level correspondence. The
final prediction results are obtained through inter-scale mixing and feature skip connection. The
experimental results indicate that MLDAC obtains 55.1% and 26.8% one-shot mIoU self-supervised
few-shot segmentation on the PASCAL-5
i
and COCO-20
i
datasets, respectively. In addition, it
achieves 78.1% on the FSS-1000 few-shot dataset, proving its efficacy.
Keywords: scene understanding; self-supervised learning; few-shot semantic segmentation;
self-supervised learning; multi-task learning; Swin Transformer
1. Introduction
With the rapid improvement of computing power (supported by advanced hardware
platforms) and the rise of deep learning algorithms, various scene awareness schemes have
been integrated into smart cars. For instance, Light Detection and Ranging (LiDAR), Radio
Detection and Ranging (RADAR) are 2 important auxiliary equipment for scene under-
standing under poor lighting conditions via radio waves and laser pulses, respectively [
1
].
They are highly valued for precise distance measurement (for scene reconstruction and
mapping), but behave poor in visual recognition aspect. Ultrasonic sensors are specially
designed for parking assistance, but they are limited to short distance warning. By con-
trast, vision sensors with the state-of-the-art recognition and perception algorithms greatly
improve the scene awareness ability of the intelligent vehicles, which also have lower
prices than the abovementioned instruments. The commonly involved algorithms in scene
understanding include object detection, semantic segmentation and depth recovery [
2
] etc.
Traditional deep learning approaches often require a massive number of labeled sam-
ples. Computer vision tasks like image segmentation typically need numerous high-quality
pixel-level annotations to guide model training [
3
,
4
]. However, acquiring such anno-
tated data is expensive, time-consuming, or even infeasible. Few-shot learning leverages
Sensors 2024,24, 4975. https://doi.org/10.3390/s24154975 https://www.mdpi.com/journal/sensors

Sensors 2024,24, 4975 2 of 15
prior knowledge to generalize to new tasks with only a few supervised samples. There-
fore, to reduce the labeling costs of image segmentation under scenarios with limited or
scarce samples, many studies have incorporated few-shot learning into the image semantic
segmentation field.
Few-Shot Semantic Segmentation (FSS) requires less annotation data to complete pixel-
level semantic segmentation tasks, improving its generalization ability. Normally, a limited
number of images is annotated for one category, which indicates that the model must learn
intra-class features and migrate them to unseen classes. Currently, mainstream FSS includes
metric learning-based methods, parameter prediction-based methods, fine tuning-based
methods, and memory-based methods. Among them, metric learning-based
methods [5–9]
play a dominant role. The distances between feature vectors supporting images and
querying images in high-dimensional space are used [
5
] to calculate the similarity between
them so as to predict the category probability of each pixel in the image.
In this method, the amount of annotation of images without visible categories is greatly
reduced, but the annotation requirement of visible categories is still indispensable in the
training process. Therefore, it is still challenging to further reduce annotation requirements
using few-shot semantic segmentation. A two-stage unsupervised image segmentation
method was proposed by the authors of [
10
], who used the K-means clustering method to
cluster pixels in images into semantic groups to obtain significant regions with continuous
semantic information. A self-supervised FSS method based on unsupervised significance
for prototype learning was devised in [
11
] to generate training pairs from a single training
image to capture the similarity between the queried image and a specific region supporting
the image. The feature vector of the Laplacian matrix derived from the feature affinity
matrix of a self-supervised network was utilized in [
12
] to eliminate the need for extensive
annotation, effectively capturing the global representation of the object of interest in the
supporting image. As mentioned earlier, traditional methods rely heavily on manual
annotation. Although the above methods alleviate such problems to some extent under a
self-supervised learning framework, most of them do not fully utilize the different scale
features, leading to poor segmentation performance.
To solve the abovementioned problem, a self-supervised few-shot segmentation
method based on saliency segmentation is proposed in this paper. Our method is built
under a multi-task learning framework. Each saliency mask is divided into two parts, one
of which is used as a support mask in a random way, while the other part is used as a query
mask to participate in the model training of few-shot meta learning. To further enhance
the meta learning effect, multiple learning tasks are proposed after saliency segmentation
to jointly enhance the few-shot segmentation performance. At the same time, in order to
enhance the robustness of the model, noise addition and image enhancement are applied
to process the input image so as to better simulate FSS tasks. To fully utilize the multi-scale
features, a dense attention calculation mechanism is developed, which transforms the
multi-scale feature map into a multi-scale dense attention block to yield the final prediction
result via inter-scale mixing. Finally, the self-supervised few-shot semantic segmentation
method is formed, which is based on a multi-task learning scheme and dense attention
calculation.
The main contributions of this paper are summarized as follows:
1.
A self-supervised few-shot segmentation method is proposed based on a multi-task
learning paradigm. The unsupervised salient part of the image is split into two parts;
one of them is used as a support image mask for few-shot segmentation, and the other
part and the entire image are used to calculate the cross-entropy with the prediction
results to realize multi-task learning so as to improve the generalization ability;
2.
An efficient few-shot segmentation network based on dense attention computation is
proposed. Multi-scale feature extraction is carried out using Swin Transformer so as
to make full use of the multi-scale pixel-level correlation.
Experimental results obtained on three mainstream datasets show that our method
surpasses other popular methods in segmentation accuracy, demonstrating its efficacy.

Sensors 2024,24, 4975 3 of 15
2. Related Works
2.1. Few-Shot Semantic Segmentation with Fully Supervised Learning
A two-branch network with a prediction-based method was proposed in [
13
], which
consists of a conditional branch and a segmentation branch. It aims to solve a one-way
one-shot image segmentation problem by using parameter prediction-based methods to
modify the classifier weights for cross–class adaptation. Instead of relying only on support
samples, query images were also used to generate classifier weights [
14
]. Instead of directly
replacing the classifier parameters, the classifier weights were dynamically added in [
15
],
enabling the model to master both base and unseen categories.
Metric learning-based methods are the most commonly used techniques for FSS.
Among them, the methods based on prototype networks are particularly prevalent. In tra-
ditional learning-based methods [
8
,
16
], the learned prototype of a class is an approximate
estimate of the optimal prototype. Recent few-shot methods aim to obtain specific pro-
totypes of objects so as to provide relevant information. Such methods provide higher
similarity scores for query characteristics belonging to the same semantic classes as the
object instead of approximating the best prototype [
5
,
6
,
17
,
18
]. When applying the masked
average pooling operation to generate a holistic descriptor for each semantic category
in these methods, some problems arise, such as the prototype learner failing to output a
robust class representation, making it difficult to capture rich and fine-grained semantic
information only using a global feature vector due to the visual differences between sup-
port and query images. To address these issues, some subsequent approaches have sought
to generate multiple prototypes for each semantic category [
9
,
19
], and others perform
intensive matching between support and query images [
20
–
22
]. The fine tuning-based
approaches aim to use an optimization algorithm to improve the parameters of the pre-
trained segmentation network to learn unseen categories [
23
–
25
]. In memory-based FSS,
semantic information is retained to assist in segmentation of query samples and obtain
more cross-resolution information and precise segmentation results [26,27].
2.2. Self-Supervised Learning for Image Semantic Segmentation
SSL bridges supervised learning and unsupervised learning. It still requires large-
scale labeled data to obtain better performance in terms of visual features. To avoid
high-cost annotation, methods are proposed to learn general image and video features
from unlabeled data without additional supervision. Self-supervised learning aims to
obtain universal image or video features by learning large-scale unlabeled data without
any manual annotation. In image segmentation, self-supervised segmentation refers to
the automatic generation of segmentation labels for images without the need for manual
annotation in order to achieve learning of image segmentation tasks.
The self-supervised semantic segmentation model predicts a set of labels (i.e., masks
with deterministic meanings) based on the input data. Previous methods [
28
,
29
] were
based on offline pre-calculated labels, followed by model updating. As a more lightweight
method, self-training on pseudo-labels is an important way to seek high-quality supervi-
sion through high-confidence class prediction. Wen et al. [
30
] defined each class of objects
as a learnable class vector and calculated the similarity between the class vector and each
position in the image feature map. They aggregated features of the same class in the image,
then constructed positive and negative sample pairs of the aggregated class features for
comparative training and learning. Araslanov et al. [
31
] applied standard data augmen-
tation techniques such as noise, flipping, and scaling to self-supervised segmentation,
ensuring consistent semantic prediction results across different image transformations.
2.3. FSS Vision Transformers
The use of transformers was originally proposed for natural language processing tasks
due to their excellent long-range dependency modeling ability. They were later migrated
to the computer vision domain [32,33].

Sensors 2024,24, 4975 4 of 15
The combination of a vision transformer and FSS is a recently emerging topic.
Lu et al. [34]
designed a Classifier Weight Transformer (CWT) to dynamically adjust the weight of the
classifier for each query image to make better use of the support set (a limited collection
of images with corresponding annotated masks to furnish the model with exemplars of
the target classes) in the query image. However, it still follows the prototype pipeline
and, therefore, cannot fully exploit the supporting information at a fine-grained level.
A novel Cyclic Consistent Transformer (CyCTR) module was developed in [
35
] that aggre-
gates pixel-level support features into query features, focusing on providing each query
pixel with relevant information from the support image to facilitate the classification of
query pixels. DCAMA [
36
] follows the design of CyCTR, introducing full exploitation of
the support image by pairing queries and supporting multi-level pixel-level correlations
between features.
3. Method
In this section, we introduce our proposed MLDAC in detail. First, the task of self-
supervised few-shot segmentation is clarified; then, our multi-task framework is introduced
in Section 3.2. The core modules in MLDAC are described in Section 3.3.
3.1. Problem Definition
Fully supervised few-shot semantic segmentation. Traditional FSS is always based
on fully supervised learning. Specifically, given the same class of images and corresponding
mask conditions in the training set (
Dtrain
), the model aims to find the designated area
related to the mask in another image based on the images and corresponding masks in
the given test set (
Dtest
) so as to accomplish the few-shot segmentation task. This is the
meta-learning paradigm called episodic training. In real applications, both
Dtrain
and
Dtest
consist of different classes of objects, and image pairs with the same category are
selected to realize the meta-learning paradigm.
Dtrain
over class
Ctrain
has a completed
annotation mask for every image. The classes (
Ctest
) of
Dtest
have no shared classes, as is
the case for
Ctrain
(i.e.,
Ctrain∩Ctest
=
Φ
). In episodic training, each image pair contains a
duplicate image, mask, and class information, where the class information is the same,
i.e., for
(x1
,
m1
,
y1)
and
(x2
,
m2
,
y2)
,
(y1=y2)
, where
x1
and
x2
are the images,
m1
and
m2
are the ground-truth binary masks, and
y1
and
y2
are the class labels corresponding to the
mask.
Self-supervised few-shot semantic segmentation (SFSS). For the self-supervised
few-shot semantic segmentation problem,
Dtrain
consists of images without masks or labels
so that the training process cannot be implemented. To solve this problem, a new SFSS
method based on multi-task learning to build a self-supervised experimental process is
proposed. After training, the same evaluation protocol as the standard FSS can be used to
evaluate the learned meta-models for a multitude of segmentation tasks with few images.
3.2. Framework
To realize SFSS, a complete episodic training framework is constructed in this paper.
The architecture of our proposed MLDAC based on multi-task learning consists of three
inputs (query image, support image, and support mask) and one output (segmentation
result), as shown in Figure 1. The input is a single image without any annotation or class
label (
Iimage∈Dtrain
). Self-supervised learning usually uses data attributes to set unsuper-
vised tasks instead of manual annotation. Therefore, unsupervised saliency prediction
was utilized to obtain the saliency region (
Msaliency
), which depicts the arbitrary object in
the image with continuous and accurate edge information. Next,
Msaliency
is divided into
2 parts,
namely
M1
saliency
and
M2
saliency
, the former of which is used as the support mask that
is input to MLDAC, while M2
saliencyand Msaliency are used to calculate the loss as follows:
loss1=CrossEntropyLoss(M1
saliency,result);ignore index =Msaliency (1)

Sensors 2024,24, 4975 5 of 15
loss2=CrossEntropyLoss(Msal iency,result);ignore index =Φ; (2)
then,
result =MLDAC(IQuery,ISupport,M2
saliency)(3)
Equations (1) and (2) are both cross-entropy loss functions with slightly different
implementation details. For Loss 1,
M1
saliency
do not participate in the calculation of the
loss function so that the model focuses on learning the query region, weakening the
impact of the support region. Loss1 and Loss2 guide model training in an alternate way,
with probabilities set to a and 1-a, respectively.
Meanwhile, since
IQuery
and
ISupport
come from the same source, to highlight the
difference between them, we employ data augmentation techniques, including jittering,
horizontal flip-flop, rotation, and random cropping. Gaussian noise is also added before
image enhancement (i.e., the color of the selected query region is perturbed slightly to aug-
ment the diversity of the training data). The pseudo code of our proposed self-supervised
few-shot semantic segmentation framework is expressed in Algorithm 1 as follows:
Algorithm 1 FSS self-supervised framework based on multi-task learning
1: multi-task split(I mage):
2: Saliency =Unsupervised Saliency Detection(Image)
3: Saliency1, Saliency2=Split(Saliency)
4: if random.random() <athen
5: target =Saliency
6: loss =nn.CrossEntropyLoss()
7: else {}
8: target =Saliency2
9: loss =nn.CrossEntropyLoss(gnore index =Saliency1)
10: end if
11: q image,target =Augmentations(Image,target)
12: s image,s mask =Augmentations(Image +b,Saliency1)
13: result =MLDAC(q image,s image,s mask)
14: loss =loss(result,target)
Query Image
Image Split
Support Image Half Split as
Support mask
Loss1
Unsupervised
Saliency
Split into
two parts Half Split as
Query mask
Augmentations
Loss2
MLDAC
Result
As Query mask
1aliencyS
M
2aliencyS
M
alinecyS
M
Figure 1. The overall structure of the proposed self-supervised network. The unsupervised saliency
mask is segmented; one part is used for masking and support, and the other part and the entire
unsupervised salient area are used to calculate the loss function so as to guide model training.

Sensors 2024,24, 4975 6 of 15
3.3. MLDAC Network Architecture
As shown in Figure 2, our proposed MLDAC consists of the following 3 parts:
In the first part, a pre-trained feature extractor is applied to process both the query and
the support images to obtain multi-scale query and support features and support image
masks of the corresponding size;
The second part inputs the query features, support features, and support masks at
each scale into a multi-layer Dense Attention Computation Block (DACB) of the same scales
as Q, K, and V. DACB performs a multi-scale query, support features, and support mask
attention calculations;
The third part involves the aggregation of the outputs from the previous stage and the
multi-scale features. This process produces the final prediction masks using a tailored mixer.
Query Image
Support Image
Multi-task
Swin-B
Conv
Upsample
Conv
Upsample
Conv
Conv Conv
Concatenate
Upsample
Concatenate
Conv
Conv & ReLU
Conv
Upsample
Conv & ReLU
Conv
Upsample
Conv & ReLU
Conv
Result
Swin-B
Shared weight
Downsample
Query
Unsupervised
Saliency
Split into
two parts
aliencyS
M
2aliencyS
M
1aliency as Support Mask
S
M
image
I
2aliencyS
M
aliencyS
M
Upsample
Skip Connection
DACB
DACB
DACB
Figure 2. The architecture of our network with the proposed self-supervised meta-learning approach.
3.3.1. Feature Extraction and Masking
The first stage involves the extraction of different levels of semantic features. Here,
Swin Transformer (Swin-B) is employed as the feature extractor, which captures both
local fine-grained features and contextual semantic information. Through the bottom-up
pathway, features at multiple scales are computed, enabling multi-scale feature learning.
Following [
7
], after capturing image features of different sizes, the image mask is scaled to
different support mask sizes via linear interpolation, allowing for cross-feature attention in
different layers. Compared with existing FSS models, the Swin Transformer model (Swin-B)
is adapted to extract features and was pre-trained on ImageNet-1K and liu2021swin.
3.3.2. Dense Attention Computation Block(DACB)
Our proposed DACB aggregates multi-scale features to produce semantic information.
The initial stage involves the transformer architecture (i.e., scaled dot-product attention).
The corresponding calculation is written as follows:
AttnQ′,K′,V=so f tm ax Q′K′T
√d!V, (4)

Sensors 2024,24, 4975 7 of 15
where
Q
,
K
, and
V
are the sets of query, key, and value vectors, respectively; d represents
the dimension of the query and key vectors; and
Q′
,
K′
indicates that the location code has
been added to Q,Kwith absolute learnable position encoding.
In this paper, the query and support feature maps are denoted as
Fq
,
Fs∈Rh×w×c
,
where
h
,
w
, and
c
represent the height, width, and number of channels of the feature maps,
respectively. As shown in Figure 3, the support feature (Fq) and query feature (Fs) are
flattened first, and each pixel value is regarded as a token. Then, after adding learnable
linear position coding, the
Q′
,
K′
matrix is generated from the flattened
Fq
and
Fs
, and the
multi-head attention mechanism is implemented as follows:
MH AQ′,K′,V=[head1,head2,· · · ,headn], (5)
where
headm=Atten(Q′
m,K′
m,Vm)
, and the inputs
[Q′
m,K′
m,Vm]
are the
mth
group from
[Q′
m,K′
m,Vm]
with dimension
d/h
. For the support mask, it is only necessary to flatten it to
participate in the calculation of dense attention. Finally, the output of the multiple attention
heads of each token is averaged, and the averaged output is reset to a two-dimensional
tensor expressed as
ˆ
R
with dimensions of
h×w×c
, which is the final dense attention
computation output.
absolute learnable position encoding
Query feature
[h, w, c] Support feature
[h, w, c]
Fatten
[h*w, c]
Linear(c, c)
[h*w, c]
Support mask
[h, w, 1]
Fatten
[h*w, 1]
Reshape
[heads, h*w, c/heads] Repeat
[heads, h*w, 1]
Q
[heads, h*w, c/heads] K
[heads, h*w, c/heads] V
[heads, h*w, 1]
Matmul
[heads, h*w, 1]
Mean and Reshape
Output
[h, w, 1]
Attention weight 𝑄 ∙ 𝐾 𝑇
[heads, h*w, h*w]
Figure 3. Illustration of the proposed DACB.
3.3.3. Inter-Scale Mixing and Up-Sampling Module
After cross-feature dense attention computation at different scales of features from
multiple layers, it is necessary to mix attention results from these different scales to obtain
the final prediction. Our inter-scale mixing and up-sampling module has 2 parts; one
stitches the different layers directly after cross-feature dense attention computation, and the
other one improves the recognition of image features using skip linking. In this step,
the size of each layer is adjusted via continuous up-sampling operations.
First, the dense attention computation scale of
1
8
,
1
16
,
1
32
is specifically used to obtain
the attention-weighted result, and
R1/8
,
R1/16
,
R1/32
.
R1/i
are subsequently processed
through several convolution blocks that are finally merged into the resultant block after

Sensors 2024,24, 4975 8 of 15
suitable up-sampling operations. The outputs of 1/32 and 1/16 are connected, resized, and
concatenated to yield the outputs of the 1/8 scale as follows:
R′
1/32+1/16 =↑k∗R′
1/32⊕k∗R′
1/16 (6)
R′=↑k∗R′
1/32+1/16 ⊕k∗R′
1/8 (7)
where
↑
is the up-sampling operation and
⊕
stands for the connection operation.
R′
is then
processed by a skip connection and decoding operation to obtain the final predicted mask.
Then, the last layer of features extracted by the feature extractor with 1/4 and 1/8 scales
are concatenated as follows:
R′′ =R′⊕Fq
1/8 ⊕Fs
1/8 (8)
F1/8+1/4 =↑Fq
1/8 ⊕ ↑ Fs
1/8 ⊕Fq
1/4 ⊕Fs
1/4 (9)
R′′′ =↑R′′ ⊕(k∗F1/8+1/4)(10)
Finally, the result is obtained using a decoder (
f(X)
) to produce the final mask predic-
tion as follows:
Mresul t =fR′′′(11)
The decoder is composed of several convolutional modules and ReLU blocks that
operate alternately, along with up-sampling operations to attain the final segmentation
resolution. The decoder blocks gradually reduce the dimensions of the output channels
to 2 (1 for foreground and the other for background) in one-way segmentation. Two
interleaved up-sampling operations are used to restore the output size to match that of the
input images.
4. Experiments and Results
Datasets. To validate the effectiveness of our proposed method, extensive experiments
were conducted on the PASCAL-5i, COCO-20iand FSS-1000 datasets.
PASCAL-5
i
is built upon PASCAL VOC [
37
]. It has 20 categories that are further
divided into 4 folds, namely 5
0
, 5
1
, 5
2
, and 5
3
. Each fold has different kinds of categories.
For instance, 5
0
includes planes, bikes, birds, etc., while 5
1
includes buses, cars, chairs, etc.
During the training for each fold, the other three folds are used as the training dataset. We
need only image data for our unsupervised training, without the tasks or class information
associated with the images. Hence, we use images from all the folds to support our training
and use the unsupervised saliency map of all folds and the folded images to assess the
average concurrency ratios of each fold and preserve the best outcomes.
Similar to PASCAL-5
i
, COCO-20
i
is derived from MS COCO [
38
], which consists of
more than 120,000 images from 80 categories. It is split into four folds denoted by 20
0
, 20
1
,
202, and 203, each of which contains 20 categories.
The FSS-1000 dataset [
39
] is set up with well-established categories; we use only
images from pre-trained categories as support and do not use images from the target
categories as part of the training set. For all datasets, the mean intersection over union
(mIoU) is used, and one-shot segmentation results are reported and compared.
4.1. Implementation Details
All experiments were performed using PyTorch framework. The pre-trained Swin-
B-based model is used as the backbone feature extractor, (which is trained on ImageNet-
1K [
33
]). Both support and query images have input sizes of 384
×
384 pixels. For optimiza-
tion, rgw Adam optimizer was applied with a learning rate of 10
−4
, a weight decay of 10
−5
,
and pixel cross-entropy loss. Each model was trained on two 3090 GPUs for 100 epochs
using the PASCAL dataset and 30 epochs using the COCO dataset, with a batch size of 16.

Sensors 2024,24, 4975 9 of 15
4.2. Comparison with Other Popular Methods
Comparisons are made in Tables 1and 2between our method and other state-of-the-art
supervised few-shot segmentation approaches and self-supervised semantic segmentation
approaches. Here, avg represents the mean intersection over union, 5
i
represents the
average category segmentation accuracy of all categories in the i-th fold, and FSS-1000
represents the segmentation accuracy on the FSS-1000 dataset. The supervised models
utilized the ground-truth segmentation mask during usual fold-based training, whereas the
unsupervised models were trained on a training set without the ground truth. As shown
in Table 1, we achieved the best results among all the self-supervised methods and even
surpassed two of the fully supervised methods. Similarly for COCO and FSS-1000, we also
achieved the best overall results among all the self-supervised methods, exceeding two of
the fully supervised methods (on COCO).
Table 1. Comparison of results on PASCAL-5ibetween our method and other popular methods.
Method 50515253avg
Supervised approaches
CWT [34] 56.9 65.2 61.2 48.8 58.0
DAN [21] 54.7 68.6 57.8 51.6 58.2
MLC [40] 60.8 71.3 61.5 56.9 62.6
HSNet [7] 67.3 72.3 62.0 63.1 66.2
CyCTR [35] 69.3 72.7 56.5 58.6 64.3
IPMT [41] 71.6 73.5 58.0 61.2 66.1
MIANet [42] 68.5 75.8 67.5 63.2 68.7
Self-supervised approaches
Saliency * [10] 51.5 49.1 48.1 39.0 46.9
MaskContrast * [10] 53.6 50.7 50.7 39.9 48.7
IPMT * [41] 57.9 57.2 55.4 43.9 53.6
MIANet * [42] 57.2 56.8 55.9 45.2 53.8
MaskSplit [11] 54.1 57.1 54.8 46.1 53.0
Ours 58.4 57.9 58.7 46.0 55.1
* represents the results obtained by adapting the methods used to assess the same settings.
Table 2. Comparison of results on COCO-20
i
and FSS-1000 between our method and other popu-
lar methods.
Method 200201202203avg FSS-1000
Supervised approaches
CWT [34] 30.3 36.6 30.5 32.2 32.4
DAN [21] - - - - 24.4 85.2
MLC [40] 50.2 37.8 27.1 30.4 36.4
HSNet [7] 37.2 44.1 42.4 41.3 41.2 86.5
PEFNet [43] 36.8 41.8 38.7 36.7 38.5
MIANet [42] 42.5 53.0 47.8 47.4 47.7
Self-supervised approaches
Saliency * [10] 22.7 24.3 20.4 22.2 22.4
HSNet [7] 29.3 25.6 20.5 23.0 24.6 76.1
MIANet * [42] 26.7 27.2 20.9 21.9 24.2 75.0
MaskSplit [11] 22.3 26.1 20.6 24.3 23.3 72.1 *
Ours 37.4 26.2 21.3 22.3 26.8 78.1
* represents the results obtained by adapting the methods used to assess the same settings.
Above all, the framework that we propose is highly effective. When comparing results
on the PASCAL dataset, we use the results reported in MASKSPLIT, where the Saliency*
and MaskContrast* methods are from [
10
], optimized to obtain unsupervised approaches
that represent the framework. In order to make a comprehensive comparison with the

Sensors 2024,24, 4975 10 of 15
existing supervised few-shot method, the source code provided by MIANet [
42
] is used to
re-conduct the experiment under the self-supervised settings, obtaining a result of 53.8%.
In comparison to supervised approaches, self-supervised approaches perform poorly
because they do not learn intra-class information. Despite this, our approach performed
exceptionally well, achieving a score of 55.1% on PASCAL. This score is two points higher
than the initial slicing approach, MASKSPLIT, which is very competitive.
Table 2shows a comparison between our method and other popular self-supervised
and fully-supervised few-shot segmentation methods on COCO and FSS-1000. As shown
by the results, the performance of our method was greatly enhanced compared to current
self-supervised few-shot segmentation methods, with an increase from 23.3 to 26.8 on the
COCO dataset and to 78.1 on the FSS-1000 dataset, which is a significant improvement.
We attribute the superb results on the FSS-1000 dataset to the unsupervised saliency
regions being more prominent and free of noise and to the relatively high within-class
image similarity. It is worth noting that MaskSplit surpasses our method on 5
3
in both
Tables 1and 2
. The reason is that MaskSplit masks out all the background regions of
the supported image (i.e., masked pooling) during the self-supervised training process.
This strategy is effective when facing a complex background. However, we realize better
overall results by using dense attention computation to acquire both the background and
foreground information.
To further demonstrate the advantages of our model for cross-domain few-shot seg-
mentation tasks, an additional experiment was conducted on the ISIC2018 dataset [
44
],
which contains skin lesion images and is mainly used for medical image analysis and
model training. It comprises thousands of high-resolution images of skin lesions, including
benign lesions such as moles and pigmented nevi, as well as malignant lesions such as
melanoma and basal cell carcinoma. The major advantage is that the images have already
been annotated by professionals.
Comparative results on ISIC are shown in Table 3. PATNet [
45
] proposes a few-
shot segmentation network based on pyramid anchoring transformation, which converts
domain-specific features into domain-independent features for downstream segmentation
modules to quickly adapt to unknown domains. PMNet [
46
] proposes a pixel matching
network that extracts domain-independent pixel-level dense matches and captures pixel–
pixel patch relationships in each supporting query pair using bidirectional 3D convolution.
Compared with PATNet [45] and PMNet [46], we achieved much higher accuracy.
Table 3. Comparative results on ISIC2018.
Method mIoU
PATNet [45] 41.16%
PMNet [46] 51.2%
Ours 65.2%
4.3. Analysis of the Computational Complexity
In this section, we analyze the computational complexity of MLDAC in terms of model
parameters (Params), floating-point operations (FLOPs), and inference time. Among them,
Params indicates the number of parameters that the model has (i.e., model size), and FLOPs
indicates the computation cost during inference. The inference time is the time that the
model spends to produce the segmentation results. The experiment was conducted on
2 RTX 3090 GPUs, and we adopted Swin-B as our backbone. A comparison is shown in
Table 4below. Compared with HSNet [
7
], although our method has higher computational
costs, it requires fewer iterations (i.e., much shorter inference time).

Sensors 2024,24, 4975 11 of 15
Table 4. Computational complexity.
Method FLOPs
Params
Number of
Iterations
Time in Each
Iteration
HSNet [7] 103.8 G 86.7 M 90 15 m
MLDAC (Ours) 112.0 G 96.1 M 18 15 m
4.4. Visualization Results
In this section, a comparison of visualization results is shown to demonstrate the seg-
mentation results obtained by different methods. As shown in The first column in Figure 4
shows the ground-truth segmentation results; the second column shows the supporting im-
age and corresponding masks; and the third and the fourth columns show the segmentation
results obtained by MaskSplit and our method, respectively. In comparison, our proposed
method clearly delineates the boundary of each object better than MaskSplit. For complex
backgrounds, MLDAC can better distinguish the foreground and the background based on
the supporting images.
Figure 4. Comparison of visualization results on PASCAL-5
i
. Columns correspond to the query
image with mask, support image with mask, MaskSplit results, and our results.

Sensors 2024,24, 4975 12 of 15
4.5. Ablation Study
A comprehensive ablation study was conducted on PASCAL to validate the effective-
ness of our proposed method.
4.5.1. Multi-Task Learning Parameter Settings
In this section, the proposed multi-task learning parameter (a) and noise injection
parameter (b) in MLDAC are described. Here,
a
represents the probability of selecting task
1 or task 2 during training, which balances the proportion of the two few-shot segmentation
target loss functions. When it is offset to 0 or 1, the network degenerates into an ordinary
single-target structure. The experimental results obtained using different values of a are
shown in Figure 5(1). It can be seen that when a = 0.15, the model achieves the optimal
result; therefore, we set a = 0.15 for all other experiments. Similarly, as shown in Figure 5(2),
the best result is reached when
b=
1. Here,
b
represents the mean of Gaussian noise mixed
into the image with additive noise.
Figure 5. Ablation experiments on the value of parameters a and b.
4.5.2. The Architecture of MLDAC
As shown in Table 5, the combination of different schemes was validated to search
for the optimal settings of MLDAC. As shown in Table 5, our proposed learnable linear
positional encoding and skip connection are, indeed, effective. The former enhances the
connections between different features, and the latter strengthens the semantic informa-
tion, making it easier to obtain the correlated region between the supporting and query
images. Meanwhile, the 1/4 and 1/8 features accomplish the segmentation task in a more
effective way.
Table 5. Ablation study on different layers of DACB.
1
81
16 1
32 Results
✓49.2
✓ ✓ 53.5
✓ ✓ ✓ 55.1
It can be seen from the data in Table 5that multi-scale DACBcan better capture
semantic information of different scales and obtain better segmentation effects than the
compared models. Removing 18 scale-intensive attention computing blocks reduces the
model performance by 0.6%, and consecutive removal of
1
8
and
1
16
level dense attention
computation blocks reduces the model’s performance.
4.5.3. Configuration of Learnable Absolute PE and Dense Skip Connections
Ablation studies were performed on the absolute learnable positional encodings and
skip connections, and the results are shown in Table 6. The absolute learnable position
encoding we use is added only at the
1
16
and
1
32
levels of dense attention computation. The

Sensors 2024,24, 4975 13 of 15
1
8
level uses the encodings fixed by sine and cosine functions with different frequencies
to save training costs. Our skip connections of the
1
4
scale refer to the up-sampling of
features from the previous layer when using features from the later layer and performing
the original skip-linking operation by using a conv module to resize the spliced features
back to the original size. Ablation experiments show that absolute learnable position
encoding facilitates the experiments. Using the dense skip connection operation only on
1
8
and
1
4
can improve the final segmentation by identifying the features in the intermediate
layer more efficiently. The combination of
1
4
scale features and
1
8
scale features with conv
blocks before skip connections can complete segmentation tasks more efficiently.
Table 6. Ablation experiments on the effectiveness of the proposed method on PASCAL-5i.
Fixed Learnable
PE
1
4Connection 1
8Connection 1
4
+
1
8
Connection
Results
✓ ✓ 54.6
✓ ✓ 54.3
✓ ✓ 54.1
✓ ✓ 53.9
✓ ✓ 54.5
✓ ✓ ✓ 55.1
✓✓✓ 54.8
5. Conclusions
Self-supervised methods have begun to prevail in multiple computer vision tasks,
including semantic segmentation. In this paper, a self-supervised few-shot segmentation
method is proposed based on multi-task learning and dense attention computation. Our
method utilizes unsupervised saliency regions for self-supervised learning for few-shot
segmentation (FSS), which avoids the need for extensive manual annotation. The unsuper-
vised saliency regions provide continuous semantic information to improve the training of
self-supervised FSS. The self-supervised FSS method based on multi-task learning is pro-
posed to solve the lack of category information, which divides the salient regions into query
regions and support regions. The introduction of an attention mechanism improves the seg-
mentation accuracy of the model. Extensive experiments were conducted on COCO-20
i
and
PASCAL-5
i
, on which our model achieved 55.1% and 26.8% one-shot mIoU, respectively.
In addition, it realized 78.1% on FSS-1000.
Despite the appealing results we achieved, the proposed self-supervised FSS method
based on saliency segmentation still cannot effectively provide continuous salient re-
gions for objects of the same category. In the future, we plan to introduce an image
generation scheme to construct a meta-learning paradigm for FSS so as to achieve higher
segmentation accuracies.
Author Contributions: Conceptualization, K.Y.; Methodology, K.Y.; Software, K.Y.; Validation, K.Y.
and Y.Z.; Formal analysis, K.Y. and W.W.; Investigation, W.W.; Resources, Y.Z.; Writing—original
draft preparation, K.Y. and W.W.; Writing—review and editing, W.W. and Y.Z.; Visualization, K.Y. All
authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by the Intelligent Policing Key Laboratory of Sichuan Province,
No. ZNJW2024KFMS004.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: The data presented in this study are available upon request from the
corresponding author. The data are not publicly available due to privacy concerns.
Conflicts of Interest: The authors declare no conflicts of interest.

Sensors 2024,24, 4975 14 of 15
References
1.
Kim, M.Y.; Kim, S.; Lee, B.; Kim, J. Enhancing Deep Learning-Based Segmentation Accuracy through Intensity Rendering and 3D
Point Interpolation Techniques to Mitigate Sensor Variability. Sensors 2024,24, 4475. [CrossRef]
2.
Jun, W.; Yoo, J.; Lee, S. Synthetic Data Enhancement and Network Compression Technology of Monocular Depth Estimation for
Real-Time Autonomous Driving System. Sensors 2024,24, 4205. [CrossRef]
3.
You, L.; Zhu, R.; Kwan, M.; Chen, M.; Zhang, F.; Yang, B.; Wong, M.; Qin, Z. Unraveling adaptive changes in electric vehicle
charging behavior toward the postpandemic era by federated meta-learning. Innovation 2024,5. [CrossRef] [PubMed]
4.
Liu, S.; You, L.; Zhu, R.; Liu, B.; Liu, R.; Yu, H.; Yuen, C. AFM3D: An Asynchronous Federated Meta-Learning Framework for
Driver Distraction Detection. In IEEE Transactions on Intelligent Transportation Systems; IEEE: Piscataway, NJ, USA, 2024.
5.
Wang, K.; Liew, J.; Zou, Y.; Zhou, D.; Feng, J. Panet: Few-shot image semantic segmentation with prototype alignment. In
Proceedings of the IEEE/CVF International Conference on Computer Vision, Seoul, Republic of Korea, 27 October–2 November
2019; pp. 9197–9206.
6.
Zhang, C.; Lin, G.; Liu, F.; Yao, R.; Shen, C. Canet: Class-agnostic segmentation networks with iterative refinement and attentive
few-shot learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, CA,
USA, 15–20 June 2019; pp. 5217–5226.
7.
Min, J.; Kang, D.; Cho, M. Hypercorrelation squeeze for few-shot segmentation. In Proceedings of the IEEE/CVF International
Conference on Computer Vision, Montreal, BC, Canada, 11–17 October 2021; pp. 6941–6952.
8.
Zhou, T.; Wang, W.; Konukoglu, E.; Van Gool, L. Rethinking semantic segmentation: A prototype view. In Proceedings of the
IEEE/CVF Conference on Computer Vision and Pattern Recognition, New Orleans, LA, USA, 18–24 June 2022; pp. 2582–2593.
9.
Yang, B.; Wan, F.; Liu, C.; Li, B.; Ji, X.; Ye, Q. Part-based semantic transform for few-shot semantic segmentation. IEEE Trans.
Neural Netw. Learn. Syst. 2021,33, 7141–7152. [CrossRef] [PubMed]
10.
Van Gansbeke, W.; Vandenhende, S.; Georgoulis, S.; Van Gool, L. Unsupervised semantic segmentation by contrasting object
mask proposals. In Proceedings of the IEEE/CVF International Conference on Computer Vision, Montreal, BC, Canada, 11–17
October 2021; pp. 10052–10062
11.
Amac, M.; Sencan, A.; Baran, B.; Ikizler-Cinbis, N.; Cinbis, R. MaskSplit: Self-supervised meta-learning for few-shot semantic
segmentation. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, Waikoloa, HI, USA, 3–8
January 2022; pp. 1067–1077.
12.
Karimijafarbigloo, S.; Azad, R.; Merhof, D. Self-supervised few-shot learning for semantic segmentation: An annotation-free
approach. arXiv 2023, arXiv:2307.14446.
13. Shaban, A.; Bansal, S.; Liu, Z.; Essa, I.; Boots, B. One-shot learning for semantic segmentation. arXiv 2017, arXiv:1709.03410
14.
Zhuge, Y.; Shen, C. Deep reasoning network for few-shot semantic segmentation. In Proceedings of the 29th ACM International
Conference on Multimedia, Virtual, 20–24 October 2021; pp. 5344–5352.
15.
Liu, L.; Cao, J.; Liu, M.; Guo, Y.; Chen, Q.; Tan, M. Dynamic extension nets for few-shot semantic segmentation. In Proceedings of
the 28th ACM International Conference on Multimedia, Seattle, WA, USA, 12–16 October 2020; pp. 1441–1449.
16.
Wang, W.; Zhou, T.; Yu, F.; Dai, J.; Konukoglu, E.; Van Gool, L. Exploring cross-image pixel contrast for semantic segmentation.
In Proceedings of the IEEE/CVF International Conference on Computer Vision, Montreal, BC, Canada, 11–17 October 2021;
pp. 7303–7313.
17. Dong, N.; Xing, E. Few-shot semantic segmentation with prototype learning. BMVC 2018,3, 4.
18.
Lang, C.; Cheng, G.; Tu, B.; Han, J. Learning what not to segment: A new perspective on few-shot segmentation. In Proceedings of
the IEEE/CVF Conference on Computer Vision and Pattern Recognition, New Orleans, LA, USA, 18–24 June 2022;
pp. 8057–8067
.
19.
Zhang, X.; Wei, Y.; Li, Z.; Yan, C.; Yang, Y. Rich embedding features for one-shot semantic segmentation. IEEE Trans. Neural Netw.
Learn. Syst. 2021,33, 6484–6493. [CrossRef] [PubMed]
20.
Zhang, C.; Lin, G.; Liu, F.; Guo, J.; Wu, Q.; Yao, R. Pyramid graph networks with connection attentions for region-based one-shot
semantic segmentation. In Proceedings of the IEEE/CVF International Conference on Computer Vision, Seoul, Republic of Korea,
27 October–2 November 2019; pp. 9587–9595 .
21.
Wang, H.; Zhang, X.; Hu, Y.; Yang, Y.; Cao, X.; Zhen, X. Few-shot semantic segmentation with democratic attention networks. In
Proceedings, Part XIII 16, Computer Vision-ECCV 2020: 16th European Conference, Glasgow, UK, 23–28 August 2020; Springer: Cham,
Switzerland, 2020; pp. 730–746.
22.
Liu, B.; Jiao, J.; Ye, Q. Harmonic feature activation for few-shot semantic segmentation. IEEE Trans. Image Process. 2021,30,
3142–3153. [CrossRef] [PubMed]
23.
Yang, X.; Wang, B.; Chen, K.; Zhou, X.; Yi, S.; Ouyang, W.; Zhou, L. Brinet: Towards bridging the intra-class and inter-class gaps
in one-shot segmentation. arXiv 2020, arXiv:2008.06226.
24.
Tian, P.; Wu, Z.; Qi, L.; Wang, L.; Shi, Y.; Gao, Y. Differentiable meta-learning model for few-shot semantic segmentation. Proc.
Aaai Conf. Artif. Intell. 2020,34, 12087–12094. [CrossRef]
25.
Boudiaf, M.; Kervadec, H.; Masud, Z.; Piantanida, P.; Ben Ayed, I.; Dolz, J. Few-shot segmentation without meta-learning: A
good transductive inference is all you need? In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern
Recognition, Nashville, TN, USA, 20–25 June 2021; pp. 13979–13988.
26.
Wu, Z.; Shi, X.; Lin, G.; Cai, J. Learning meta-class memory for few-shot semantic segmentation. In Proceedings of the IEEE/CVF
International Conference on Computer Vision, Montreal, BC, Canada, 11–17 October 2021; pp. 517–526.

Sensors 2024,24, 4975 15 of 15
27.
Xie, G.; Xiong, H.; Liu, J.; Yao, Y.; Shao, L. Few-shot semantic segmentation with cyclic memory network. In Proceedings of the
IEEE/CVF International Conference on Computer Vision, Montreal, BC, Canada, 11–17 October 2021; pp. 7293–7302.
28.
Li, G.; Kang, G.; Liu, W.; Wei, Y.; Yang, Y. Content-consistent matching for domain adaptive semantic segmentation. In European
Conference on Computer Vision; Springer International Publishing: Cham, Switzerland, 2020; pp. 440–456.
29.
Subhani, M.; Ali, M. Learning from scale-invariant examples for domain adaptation in semantic segmentation. In Proceedings,
Part XXII 16, Computer Vision-ECCV 2020: 16th European Conference, Glasgow, UK, 23–28 August 2020; Springer: Cham, Switzerland,
2020; pp. 290–306.
30.
Wen, X.; Zhao, B.; Zheng, A.; Zhang, X.; Qi, X. Self-supervised visual representation learning with semantic grouping. Adv. Neural
Inf. Process. Syst. 2022,35, 16423–16438.
31.
Araslanov, N.; Roth, S. Self-supervised augmentation consistency for adapting semantic segmentation. In Proceedings of the
IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, TN, USA, 20–25 June 2021; pp. 15384–15394.
32.
Dosovitskiy, A.; Beyer, L.; Kolesnikov, A.; Weissenborn, D.; Zhai, X.; Unterthiner, T.; Dehghani, M.; Minderer, M.; Heigold, G.;
Gelly, S. Others An image is worth 16 ×16 words: Transformers for image recognition at scale. arXiv 2020, arXiv:2010.11929.
33.
Liu, Z.; Lin, Y.; Cao, Y.; Hu, H.; Wei, Y.; Zhang, Z.; Lin, S.; Guo, B. Swin transformer: Hierarchical vision transformer using shifted
windows. In Proceedings of the IEEE/CVF International Conference on Computer Vision, Montreal, BC, Canada, 11–17 October
2021; pp. 10012–10022.
34.
Lu, Z.; He, S.; Zhu, X.; Zhang, L.; Song, Y.; Xiang, T. Simpler is better: Few-shot semantic segmentation with classifier weight
transformer. In Proceedings of the IEEE/CVF International Conference on Computer Vision, Montreal, BC, Canada, 11–17
October 2021; pp. 8741–8750.
35.
Zhang, G.; Kang, G.; Yang, Y.; Wei, Y. Few-shot segmentation via cycle-consistent transformer. Adv. Neural Inf. Process. Syst. 2021,
34, 21984–21996.
36.
Shi, X.; Wei, D.; Zhang, Y.; Lu, D.; Ning, M.; Chen, J.; Ma, K.; Zheng, Y. Dense cross-query-and-support attention weighted mask
aggregation for few-shot segmentation. In European Conference on Computer Vision; Springer Nature: Cham, Switzerland, 2022;
pp. 151–168.
37.
Everingham, M.; Van Gool, L.; Williams, C.; Winn, J.; Zisserman, A. The pascal visual object classes (voc) challenge. Int. J. Comput.
Vis. 2010,88, 303–338. [CrossRef]
38.
Lin, T.; Maire, M.; Belongie, S.; Hays, J.; Perona, P.; Ramanan, D.; Dollar, P.; Zitnick, C. Microsoft coco: Common objects in context.
In Proceedings, Part V 13, Computer Vision-ECCV 2014: 13th European Conference, Zurich, Switzerland, 6–12 September 2014; Springer:
Cham, Switzerland, 2014; pp. 740–755.
39.
Li, X.; Wei, T.; Chen, Y.; Tai, Y.; Tang, C. Fss-1000: A 1000-class dataset for few-shot segmentation. In Proceedings of the IEEE/CVF
Conference on Computer Vision and Pattern Recognition, Seattle, WA, USA, 13–19 June 2020; pp. 2869–2878.
40.
Yang, L.; Zhuo, W.; Qi, L.; Shi, Y.; Gao, Y. Mining latent classes for few-shot segmentation. In Proceedings of the IEEE/CVF
International Conference on Computer Vision, Montreal, BC, Canada, 11–17 October 2021; pp. 8721–8730.
41.
Liu, Y.; Liu, N.; Yao, X.; Han, J. Intermediate prototype mining transformer for few-shot semantic segmentation. Adv. Neural Inf.
Process. Syst. 2022,35, 38020–38031.
42.
Yang, Y.; Chen, Q.; Feng, Y.; Huang, T. MIANet: Aggregating Unbiased Instance and General Information for Few-Shot Semantic
Segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Vancouver, BC,
Canada, 17–24 June 2023; pp. 7131–7140.
43.
Tian, Z.; Zhao, H.; Shu, M.; Yang, Z.; Li, R.; Jia, J. Prior guided feature enrichment network for few-shot segmentation. IEEE Trans.
Pattern Anal. Mach. Intell. 2020,44, 1050–1065. [CrossRef] [PubMed]
44.
Codella, N.; Rotemberg, V.; Tsch, L.P.; Celebi, M.; Dusza, S.; Gutman, D.; Helba, B.; Kalloo, A.; Liopyris, K.; Marchetti, M. Others
Skin lesion analysis toward melanoma detection 2018: A challenge hosted by the international skin imaging collaboration (isic).
arXiv 2019, arXiv:1902.03368.
45.
Lei, S.; Zhang, X.; He, J.; Chen, F.; Du, B.; Lu, C. Cross-domain few-shot semantic segmentation. In European Conference on
Computer Vision; Springer Nature: Cham, Switzerland, 2022; pp. 73–90.
46.
Chen, H.; Dong, Y.; Lu, Z.; Yu, Y.; Han, J. Pixel Matching Network for Cross-Domain Few-Shot Segmentation. In Proceedings of
the IEEE/CVF Winter Conference on Applications of Computer Vision, Waikoloa, HI, USA, 3–8 January 2024; pp. 978–987.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.