We present a deep learning-based multitask framework for joint 3D human pose estimation and action recognition from RGB sensors using simple cameras. The approach proceeds along two stages. In the first, a real-time 2D pose detector is run to determine the precise pixel location of important keypoints of the human body. A two-stream deep neural network is then designed and trained to map detected 2D keypoints into 3D poses. In the second stage, the Efficient Neural Architecture Search (ENAS) algorithm is deployed to find an optimal network architecture that is used for modeling the spatio-temporal evolution of the estimated 3D poses via an image-based intermediate representation and performing action recognition. Experiments on Human3.6M, MSR Action3D and SBU Kinect Interaction datasets verify the effectiveness of the proposed method on the targeted tasks. Moreover, we show that the method requires a low computational budget for training and inference. In particular, the experimental results show that by using a monocular RGB sensor, we can develop a 3D pose estimation and human action recognition approach that reaches the performance of RGB-depth sensors. This opens up many opportunities for leveraging RGB cameras (which are much cheaper than depth cameras and extensively deployed in private and public places) to build intelligent recognition systems.

Designing motion representations for 3D human action recognition from skeleton sequences is an important yet challenging task. An effective representation should be robust to noise, invariant to viewpoint changes and result in a good performance with low-computational demand. Two main challenges in this task include how to efficiently represent spatio–temporal patterns of skeletal movements and how to learn their discriminative features for classification tasks. This paper presents a novel skeleton-based representation and a deep learning framework for 3D action recognition using RGB-D sensors. We propose to build an action map called SPMF (Skeleton Posture-Motion Feature), which is a compact image representation built from skeleton poses and their motions. An Adaptive Histogram Equalization (AHE) algorithm is then applied on the SPMF to enhance their local patterns and form an enhanced action map, namely Enhanced-SPMF. For learning and classification tasks, we exploit Deep Convolutional Neural Networks based on the DenseNet architecture to learn directly an end-to-end mapping between input skeleton sequences and their action labels via the Enhanced-SPMFs. The proposed method is evaluated on four challenging benchmark datasets, including both individual actions, interactions, multiview and large-scale datasets. The experimental results demonstrate that the proposed method outperforms previous state-of-the-art approaches on all benchmark tasks, whilst requiring low computational time for training and inference.

Methods based on 64-beam LiDAR can provide very precise 3D object detection. However, highly accurate LiDAR sensors are extremely costly: a 64-beam model can cost approximately USD 75,000. We previously proposed SLS–Fusion (sparse LiDAR and stereo fusion) to fuse low-cost four-beam LiDAR with stereo cameras that outperform most advanced stereo–LiDAR fusion methods. In this paper, and according to the number of LiDAR beams used, we analyzed how the stereo and LiDAR sensors contributed to the performance of the SLS–Fusion model for 3D object detection. Data coming from the stereo camera play a significant role in the fusion model. However, it is necessary to quantify this contribution and identify the variations in such a contribution with respect to the number of LiDAR beams used inside the model. Thus, to evaluate the roles of the parts of the SLS–Fusion network that represent LiDAR and stereo camera architectures, we propose dividing the model into two independent decoder networks. The results of this study show that—starting from four beams—increasing the number of LiDAR beams has no significant impact on the SLS–Fusion performance. The presented results can guide the design decisions by practitioners.

Today, the popularity of self-driving cars is growing at an exponential rate and is starting to creep onto the roads of developing countries. For autonomous vehicles to function, one of the essential features that needs to be developed is the ability to perceive their surroundings. To do this, sensors such as cameras, LiDAR, or radar are integrated to collect raw data. The objective of this paper is to evaluate a fusion solution of cameras and LiDARs (4 and 64 beams) for 3D object detection in foggy weather conditions. The data from the two input sensors are fused and an analysis of the contribution of each sensor on its own is then performed. In our analysis, we calculate average precision using the popular KITTI dataset, on which we have applied different intensities of fog (on a dataset we have called Multifog KITTI). The main results observed are as follows. Performances with stereo camera and 4 or 64 beams LiDAR are high (90.15%, 89.26%). Performance of the 4 beams LiDAR alone decreases sharply in foggy weather conditions (13.43%). Performance when using only a camera-based model remains quite high (89.36%). In conclusion, stereo cameras on their own are capable of detecting 3D objects in foggy weather with high accuracy and their performance slightly improves when used in conjunction with LIDAR sensors.

The increasingly present automation of vehicles generates great concern both in the industrial and public sectors. Evidently, the autonomy of vehicles is accompanied by a large number of scientific questions: what are the real benefits of automating transport services? what are the new expected situations due to the presence of an autonomous vehicle? etc. In terms of road safety, great interest has been aroused in the interactions between autonomous vehicles and the rest of the road users (pedestrians, cyclists, motorists, etc.). Especially, there is interest in the way current road users act/react to the inclusion of this new element (the autonomous vehicle) in the road ecosystem. The definition and classification of these interactions undoubtedly influence the Operational Design Domain (ODD) as well as the defined behaviour that the autonomous vehicle must perform in any situation. In this article, the description of the methodology carried out for the identification, registration and subsequent analysis of these situations of interest is presented, which is composed of five parts: 1) Installation of fixed cameras and GPS system used for monitoring interactions, 2) detection of the autonomous vehicle (AV) and other road users, 3) estimation of AV and other users trajectories, 4) estimation of the distances between the AV and other users, and 5) interpretation of the data collected and its use to improve road safety.