Abstract
An autonomous vehicle relying on LiDAR data should be able to assess its limitations in real time without depending on external information or additional sensors. The point cloud generated by the sensor is subjected to significant degradation under adverse weather conditions (rain, fog, and snow), which limits the vehicle's visibility and performance. With this in mind, we show that point cloud data contains sufficient information to estimate the weather accurately and present MobileWeatherNet, a LiDAR-only convolutional neural network that uses the bird's-eye view 2D projection to extract point clouds' weather condition and improves state-of-the-art performance by 15% in terms of the balanced accuracy while reducing inference time by 63%. Moreover, this paper demonstrates that among common architectures, the use of the bird's eye view significantly enhances their performance without an increase in complexity. To the extent of our knowledge, this is the first approach that uses deep learning for weather estimation using point cloud data in the form of a bird's-eye-view projection.

Abstract
Early detection of melanoma is crucial for preventing severe complications and increasing the chances of successful treatment. Existing deep learning approaches for melanoma skin lesion diagnosis are deemed black-box models, as they omit the rationale behind the model prediction, compromising the trustworthiness and acceptability of these diagnostic methods. Attempts to provide concept-based explanations are based on post-hoc approaches, which depend on an additional model to derive interpretations. In this paper, we propose an inherently interpretable framework to improve the interpretability of concept-based models by incorporating a hard attention mechanism and a coherence loss term to assure the visual coherence of concept activations by the concept encoder, without requiring the supervision of additional annotations. The proposed framework explains its decision in terms of human-interpretable concepts and their respective contribution to the final prediction, as well as a visual interpretation of the locations where the concept is present in the image. Experiments on skin image datasets demonstrate that our method outperforms existing black-box and concept-based models for skin lesion classification. © 2023 IEEE.

Abstract
Concerns with the interpretability of ML models are growing as the technology is used in increasingly sensitive domains (e.g., health and public administration). Synthetic data can be used to understand models better, for instance, if the examples are generated close to the frontier between classes. However, data augmentation techniques, such as Generative Adversarial Networks (GAN), have been mostly used to generate training data that leads to better models. We propose a variation of GANs that, given a model, generates realistic data that is classified with low confidence by a given classifier. The generated examples can be used in order to gain insights on the frontier between classes. We empirically evaluate our approach on two well-known image classification benchmark datasets, MNIST and Fashion MNIST. Results show that the approach is able to generate images that are closer to the frontier when compared to the original ones, but still realistic. Manual inspection confirms that some of those images are confusing even for humans.

Abstract
Human-Robot Collaboration is a critical component of Industry 4.0, contributing to a transition towards more flexible production systems that are quickly adjustable to changing production requirements. This paper aims to increase the natural collaboration level of a robotic engine assembly station by proposing a cognitive system powered by computer vision and deep learning to interpret implicit communication cues of the operator. The proposed system, which is based on a residual convolutional neural network with 34 layers and a long -short term memory recurrent neural network (ResNet-34 + LSTM), obtains assembly context through action recognition of the tasks performed by the operator. The assembly context was then integrated in a collaborative assembly plan capable of autonomously commanding the robot tasks. The proposed model showed a great performance, achieving an accuracy of 96.65% and a temporal mean intersection over union (mIoU) of 94.11% for the action recognition of the considered assembly. Moreover, a task-oriented evaluation showed that the proposed cognitive system was able to leverage the performed human action recognition to command the adequate robot actions with near-perfect accuracy. As such, the proposed system was considered as successful at increasing the natural collaboration level of the considered assembly station.

Abstract
The problem of gathering sufficiently representative data, such as those about human actions, shapes, and facial expressions, is costly and time-consuming and also requires training robust models. This has led to the creation of techniques such as transfer learning or data augmentation. However, these are often insufficient. To address this, we propose a semi-automated mechanism that allows the generation and editing of visual scenes with synthetic humans performing various actions, with features such as background modification and manual adjustments of the 3D avatars to allow users to create data with greater variability. We also propose an evaluation methodology for assessing the results obtained using our method, which is two-fold: (i) the usage of an action classifier on the output data resulting from the mechanism and (ii) the generation of masks of the avatars and the actors to compare them through segmentation. The avatars were robust to occlusion, and their actions were recognizable and accurate to their respective input actors. The results also showed that even though the action classifier concentrates on the pose and movement of the synthetic humans, it strongly depends on contextual information to precisely recognize the actions. Generating the avatars for complex activities also proved problematic for action recognition and the clean and precise formation of the masks.

Abstract
Purpose: Visual perception enables robots to perceive the environment. Visual data is processed using computer vision algorithms that are usually time-expensive and require powerful devices to process the visual data in real-time, which is unfeasible for open-field robots with limited energy. This work benchmarks the performance of different heterogeneous platforms for object detection in real-time. This research benchmarks three architectures: embedded GPU-Graphical Processing Units (such as NVIDIA Jetson Nano 2 GB and 4 GB, and NVIDIA Jetson TX2), TPU-Tensor Processing Unit (such as Coral Dev Board TPU), and DPU-Deep Learning Processor Unit (such as in AMD-Xilinx ZCU104 Development Board, and AMD-Xilinx Kria KV260 Starter Kit). Methods: The authors used the RetinaNet ResNet-50 fine-tuned using the natural VineSet dataset. After the trained model was converted and compiled for target-specific hardware formats to improve the execution efficiency.Conclusions and Results: The platforms were assessed in terms of performance of the evaluation metrics and efficiency (time of inference). Graphical Processing Units (GPUs) were the slowest devices, running at 3 FPS to 5 FPS, and Field Programmable Gate Arrays (FPGAs) were the fastest devices, running at 14 FPS to 25 FPS. The efficiency of the Tensor Processing Unit (TPU) is irrelevant and similar to NVIDIA Jetson TX2. TPU and GPU are the most power-efficient, consuming about 5 W. The performance differences, in the evaluation metrics, across devices are irrelevant and have an F1 of about 70 % and mean Average Precision (mAP) of about 60 %.