Joel Vidal's research while affiliated with Universitat de Girona and other places

Publications (13)

Article
Full-text available
Recently, 6D pose estimation methods have shown robust performance on highly cluttered scenes and different illumination conditions. However, occlusions are still challenging, with recognition rates decreasing to less than 10% for half-visible objects in some datasets. In this paper, we propose to use top-down visual attention and color cues to boo...
Preprint
Full-text available
Real-time robotic grasping, supporting a subsequent precise object-in-hand operation task, is a priority target towards highly advanced autonomous systems. However, such an algorithm which can perform sufficiently-accurate grasping with time efficiency is yet to be found. This paper proposes a novel method with a 2-stage approach that combines a fa...
Article
Dynamic Contrast Enhanced Magnetic Resonance Imaging (DCE-MRI) has been recognized as an effective tool for Breast Cancer (BC) diagnosis. Automatic BC analysis from DCE-MRI depends on features extracted particularly from lesions, hence, lesions need to be accurately segmented as a prior step. Due to the time and experience required to manually segm...
Article
Full-text available
Real-time robotic grasping, supporting a subsequent precise object-in-hand operation task, is a priority target towards highly advanced autonomous systems. However, such an algorithm which can perform sufficiently-accurate grasping with time efficiency is yet to be found. This paper proposes a novel method with a 2-stage approach that combines a fa...
Chapter
Dynamic Contrast Enhanced Magnetic Resonance Imaging (DCE-MRI) is a popular tool for the diagnosis of breast lesions due to its effectiveness, especially in a high risk population. Accurate lesion segmentation is an important step for subsequent analysis, especially for computer aided diagnosis systems. However, manual breast lesion segmentation of...
Article
Full-text available
In manufacturing industries, offline programming (OLP) platforms provide an independent methodology for robot integration using 3D model simulation away from the actual robot cell and production process, reducing integration time and costs. However, traditional OLP platforms still require prior knowledge of the workpiece position in a predefined en...
Chapter
Full-text available
We propose a benchmark for 6D pose estimation of a rigid object from a single RGB-D input image. The training data consists of a texture-mapped 3D object model or images of the object in known 6D poses. The benchmark comprises of: (i) eight datasets in a unified format that cover different practical scenarios, including two new datasets focusing on...
Preprint
Full-text available
We propose a benchmark for 6D pose estimation of a rigid object from a single RGB-D input image. The training data consists of a texture-mapped 3D object model or images of the object in known 6D poses. The benchmark comprises of: i) eight datasets in a unified format that cover different practical scenarios, including two new datasets focusing on...
Article
Full-text available
Pose estimation of free-form objects is a crucial task towards flexible and reliable highly complex autonomous systems. Recently, methods based on range and RGB-D data have shown promising results with relatively high recognition rates and fast running times. On this line, this paper presents a feature-based method for 6D pose estimation of rigid o...
Article
The Point Pair Feature (Drost et al. 2010) has been one of the most successful 6D pose estimation method among model-based approaches as an efficient, integrated and compromise alternative to the traditional local and global pipelines. During the last years, several variations of the algorithm have been proposed. Among these extensions, the solutio...

Citations

... This observation can not only provide precise segmentation and quantitative assessments of breast cancer but also assist in image analysis including detection, feature extraction, classification, and treatment. In most previous studies, tumors were segmented manually, which are prone to inter-and intra-observer variabilities (34, 38,39). Furthermore, for the 3D medical imaging process, it is difficult and time-consuming for radiologists to measure lesions manually. ...
... In this work we propose an automated segmentation method for breast lesions in DCE-MRI. This method is based on our previous work in which a ROI guided, 3D patch based U-Net framework was proposed [19]. In this paper we improve the method by using a modified U-Net architecture that incorporates residual basic blocks. ...
... In another development, a 3D-vision based offline path programming system using spatial data from CAD models has been proposed and developed. It is the culmination of features from OLP platforms and a simplistic 3D vector space depicting IDE [31]. In another work, a sequential guidance platform for the design and programming of generic robots tasked with handling items and structures is proposed. ...
... The 6D poses are estimated from the predicted many-to-many 2D-3D correspondences by a RANSAC-based robust fitting procedure. In the BOP Challenge 2019 [100,106], the method outperformed all RGB and most RGB-D and D methods on the T-LESS [102], YCB-V [271], and LM-O [17] datasets. ...
... Early methods for 6DoF object pose estimation assumed a grayscale or RGB input image and relied on local image features [46,9] or template matching [7]. After the introduction of Kinect-like sensors, methods based on RGB-D template matching [21,26], point-pair features [15,22,78], 3D local features [19], and learning-based methods [5,72,35] demonstrated superior performance over RGB-only counterparts. Recent methods are based on convolutional neural networks (CNNs) and focus primarily on estimating the pose from RGB images. ...
... This allows to freely integrate various different sources and technologies that are commonly used to sense pose information. Examples include indoor localization systems based on WiFi, RFID, UWB or Bluetooth [6], common localization techniques for mobile robots based on laser, magnetic or optical sensors [7], vision based pose estimator that use fiducial markers [8] or even pose estimators the leverage deep neural networks [9]. Consider the situation shown in figure 1: An autonomous mobile robot (AMR) must deliver an object to a transfer unit in an unknown location. ...
... In this section, we present and discuss the experiment result of the items mentioned in Section 4 (4.1, 4.2, and 4.3). In previous works, the Point Pair Feature voting approach [22,53] was evaluated against the state-of-the-art solutions and outperformed these solutions on a set of extensive and publicly available datasets with a variety of challenging real-world scenarios under clu er and occlusion. Table 3. with the 2 metrics 5cm5deg and ADD have shown a clear advantage of our method compared to [22]. ...
... A trend in recent years is to ensure indoor location systems including self-sufficient robot management (Wang et al. 2016;Prorok and Martinoli 2011), position identification (Chen et al. 2016), and location-based services (Shin et al. 2015;Bordel et al. 2017;Ishida et al. 2016). For instance, there are several localization and positioning systems proposed in the literature, such as those based on GPS (Gowdayyanadoddi et al. 2015), RFID (Zhao et al. 2017), infrared (Vidal and Lin 2016), ultrasound (Hammoud et al. 2016), WLAN (Khalajmehrabadi et al. 2016), Bluetooth (Gu and Ren 2015), and other approaches (Yassin et al. 2016). However, GPS may not be fit-for-purpose in indoor situations due to multipath fading (e.g., caused by objects and surfaces) and power attenuation. ...