Fig 4 - uploaded by Michael Heeney
Content may be subject to copyright.
ROC curves Class Precision Recall ROC area Walkable 0.77 0.57 0.73 Non-walkable 0.55 0.76 0.71 Average 0.68 0.64 0.73
Context in source publication
Context 1
... random guess would correspond to the diagonal, and any curve above the diagonal is an improvement over the random guess, with larger areas corresponding to better classifiers. Figure 4 shows visually that the SVM is clearly superior to the other two options. Notice that in our evaluation we have considered images from different paths. ...
Similar publications
From linear classifiers to neural networks, image classification has been a widely explored topic in mathematics, and many algorithms have proven to be effective classifiers. However, the most accurate classifiers typically have significantly high storage costs, or require complicated procedures that may be computationally expensive. We present a n...
The complexity of state-of-the-art modeling techniques for image classification impedes the ability to explain model predictions in an interpretable way. A counterfactual explanation highlights the parts of an image which, when removed, would change the predicted class. Both legal scholars and data scientists are increasingly turning to counterfact...
Deep neural networks, often equipped with powerful auto-optimization tools, find widespread use in diverse domains like NLP and computer vision. However, traditional neural architectures come with specific inductive biases, designed to reduce parameter search space, cut computational costs, or introduce domain expertise into the network design. In...
The use of social media data in disaster and crisis management is increasing rapidly. Particularly in connection to flooding events, geo-referenced images shared by citizens can provide situational awareness to emergency responders, as well as assistance to financial loss assessment, giving information that would otherwise be very hard to collect t...
Citations
... OpenCV is an open-source and cross-platform computer vision library that contains many algorithms that are used in digital image processing and provides a large number of Java interfaces [40]. Thus, this study uses the OpenCV function library to extract and classify the image features. ...
An image classification algorithm based on adaptive feature weight updating is proposed to address the low classification accuracy of the current single-feature classification algorithms and simple multifeature fusion algorithms. The MapReduce parallel programming model on the Hadoop platform is used to perform an adaptive fusion of hue, local binary pattern (LBP) and scale-invariant feature transform (SIFT) features extracted from images to derive optimal combinations of weights. The support vector machine (SVM) classifier is then used to perform parallel training to obtain the optimal SVM classification model, which is then tested. The Pascal VOC 2012, Caltech 256 and SUN databases are adopted to build a massive image library. The speedup, classification accuracy and training time are tested in the experiment, and the results show that a linear growth tendency is present in the speedup of the system in a cluster environment. In consideration of the hardware costs, time, performance and accuracy, the algorithm is superior to mainstream classification algorithms, such as the power mean SVM and convolutional neural network (CNN). As the number and types of images both increase, the classification accuracy rate exceeds 95%. When the number of images reaches 80,000, the training time of the proposed algorithm is only 1/5 that of traditional single-node architecture algorithms. This result reflects the effectiveness of the algorithm, which provides a basis for the effective analysis and processing of image big data.
... The path is open even for organic extensions and stimulation of sensory axons in many different areas of life, as described e.g. in [7]. By advanced human augmentation, people may also be empowered with new cognitive abilities and direct connection to global information and services with technologies that serve as "external eyes and ears" for sensing the ubiquitous information [8,9]. Consequently, human augmentation technologies can improve the quality of life for people with special needs, including the blind, the deaf, people with motor impairments, etc. Elderly people suffering from a deterioration of the senses, people with certain illnesses, and people injured in accidents may be helped with new solutions that replace or improve damaged senses [10]. ...
Advanced human augmentation provides a human-centred perspective on technology design. It builds upon earlier technological concepts such as ubiquitous computing, wearable computing, augmented-virtual-and mixed realities, autonomous systems and ambient intelligence. This tutorial contemplates advanced human augmentation in the industrial work context, and considers the requirements for a future augmented Superworker and the prerequisites for their advanced augmentation. In this tutorial, it is anticipated that to support the design of new augmenting solutions current human-centred design practices should be reconsidered and enhanced in new directions.
Disabled People deal with a series of barriers that limit their inclusion, empowerment, well-being, and role in society with a special emphasis in low and medium-income countries. One of these barriers is concerning the accessibility and affordability of assistive technologies (ATs) that help to enhance the quality of life of these persons. In this context, this systematic literature review (SLR) analyzes and describes how free and open-source hardware (OSHW) and open software (OSS) are employed in the design, development, and deployment of low-cost ATs. In the SLR process, different ATs were analyzed for disabilities such as visual, mobility, upper body, prostheses, hearing & speaking, daily living, and participation in society. The ATs were designed with diverse OSHW and OSS technologies such as Arduino, Raspberry Pi, NVidia Jetson, OpenCV, YOLO, MobileNet, EEG and EMG signal conditioning devices, actuators, and sensors such as ultrasonic, LiDar, or flex. 809 studies were collected and analyzed from the database Web of Science, GitHub, and the specialized journals in OSHW
HardwareX
and the
Journal of Open Hardware
during the years 2013-2022. In the first part of the SLR, the bibliometric trends and topic clusters regarding the selected studies are described. Secondly, the ATs identified with open source technologies, e.g., sensor-based or computer vision-based, are described along with a complete state-of-art about these based on each disability recognized. Finally, the issues and challenges to this approach are explored including technical factors, documentation, government policies, and the inclusion of disabled people in open source co-creation. The purpose of this study is to inform practitioners, designers, or stakeholders about low-cost (frugal) ATs with OSHW and OSS, and thus promote their development, accessibility, and affordability, contributing to benefit the community of disabled people.
Building reliable surveillance systems is critical for security and safety. A core component of any surveillance system is the human detection model. With the recent advances in the hardware and embedded devices, it becomes possible to make a real-time human detection system with low cost. This paper surveys different systems and techniques that have been deployed on embedded devices such as Raspberry Pi. The characteristics of datasets, feature extraction techniques, and machine learning models are covered. A unified dataset is utilized to compare different systems with respect to accuracy and performance time. New enhancements are suggested, and future research directions are highlighted.
Deep learning is a very hot topic currently in pattern recognition and artificial intelligence researches. Aiming at the practical problem that people usually don’t know correct classifications some rubbish should belong to, based on the powerful image classification ability of the deep learning method, we have designed a prototype system to help users to classify kinds of rubbish. Firstly the CaffeNet Model was adopted for our classification network training on the ImageNet dataset, and the trained network was deployed on a web server. Secondly an android app was developed for users to capture images of unclassified rubbish, upload images to the web server for analyzing backstage and retrieve the feedback, so that users can obtain the classification guide by an android device conveniently. Tests on our prototype system of rubbish classification show that: an image of one single type of rubbish with origin shape can be better used to judge its classification, while an image containing kinds of rubbish or rubbish with changed shape may fail to help users to decide rubbish’s classification. However, the system still shows promising auxiliary function for rubbish classification if the network training strategy can be optimized further.
