Rich and complex events in sports have led to the development of a wide-variety of techniques for interpreting content of sports videos in terms of players’ actions, poses, gait, performance, etc. This is due to the requirements from coaches, trainers and players who expect to analyze actions in top sports events, as well as sports fans who practice to imitate professional playing skills, e.g., dribbling, shooting, etc. However, this poses two key challenges for automated sports analysis community. Firstly, there are extremely limited public sports datasets. Secondly, recent advances in interpretations of sports activities, e.g., soccer, are predominantly made through analyzing coarse-grained contents. Players’ fine-grained skills analysis still remains under-explored. To alleviate these problems, this paper (a) collects the dataset of highlight videos of soccer players, including two coarse-grained action types of soccer players and six fine-grained actions of players. Detailed annotations are provided for the collected dataset, in terms of action classes, bounding boxes, segmentation maps, and body keypoints of soccer players, and positions of a soccer ball in a game. (b) leverages the understanding of complex highlight videos by proposing an energy-motion features aggregation network-EMANet to fully exploit energy-based representation of soccer players movements in video sequences and explicit motion dynamics of soccer players in videos for soccer players’ fine-grained action analysis. Experimental results and ablation studies validate the proposed approach in recognizing soccer players actions using the collected soccer highlight video datasets.

Self-supervised monocular depth estimation has seen significant progress in recent years, especially in outdoor environments, i.e., autonomous driving scenes. However, depth prediction results are not satisfying in indoor scenes where most of the existing data are captured with hand-held devices. As compared to outdoor environments, estimating depth of monocular videos for indoor environments, using self-supervised methods, results in two additional challenges: (i) the depth range of indoor video sequences varies a lot across different frames, making it difficult for the depth network to induce consistent depth cues for training, whereas the maximum distance in outdoor scenes mostly stays the same as the camera usually sees the sky; (ii) the indoor sequences recorded with handheld devices often contain much more rotational motions, which cause difficulties for the pose network to predict accurate relative camera poses, while the motions of outdoor sequences are pre-dominantly translational, especially for street-scene driving datasets such as KITTI. In this work, we propose a novel framework-MonoIndoor++ by giving special considerations to those challenges and consolidating a set of good practices for improving the performance of self-supervised monocular depth estimation for indoor environments. First, a depth factorization module with transformer-based scale regression network is proposed to estimate a global depth scale factor explicitly, and the predicted scale factor can indicate the maximum depth values. Second, rather than using a single-stage pose estimation strategy as in previous methods, we propose to utilize a residual pose estimation module to estimate relative camera poses across consecutive frames iteratively. Third, to incorporate extensive coordinates guidance for our residual pose estimation module, we propose to perform coordinate convolutional encoding directly over the inputs to pose networks. The proposed method is validated on a variety of benchmark indoor datasets, i.e., EuRoC MAV, NYUv2, ScanNet and 7-Scenes, demonstrating the state-of-the-art performance. In addition, the effectiveness of each module is shown through a carefully conducted ablation study and the good generalization and universality of our trained model is also demonstrated, specifically on ScanNet and 7-Scenes datasets.

We consider the problem of estimating frame-level full human body meshes given a video of a person with natural motion dynamics. While much progress in this field has been in single image-based mesh estimation, there has been a recent uptick in efforts to infer mesh dynamics from video given its role in alleviating issues such as depth ambiguity and occlusions. However, a key limitation of existing work is the assumption that all the observed motion dynamics can be modeled using one dynamical/recurrent model. While this may work well in cases with relatively simplistic dynamics, inference with in-the-wild videos presents many challenges. In particular, it is typically the case that different body parts of a person undergo different dynamics in the video, e.g., legs may move in a way that may be dynamically different from hands (e.g., a person dancing). To address these issues, we present a new method for video mesh recovery that divides the human mesh into several local parts following the standard skeletal model. We then model the dynamics of each local part with separate recurrent models, with each model conditioned appropriately based on the known kinematic structure of the human body. This results in a structure-informed local recurrent learning architecture that can be trained in an end-to-end fashion with available annotations. We conduct a variety of experiments on standard video mesh recovery benchmark datasets such as Human3.6M, MPI-INF-3DHP, and 3DPW, demonstrating the efficacy of our design of modeling local dynamics as well as establishing state-of-the-art results based on standard evaluation metrics.

Self-supervised depth estimation for indoor environments is more challenging than its outdoor counterpart in at least the following two aspects: (i) the depth range of indoor sequences varies a lot across different frames, making it difficult for the depth network to induce consistent depth cues, whereas the maximum distance in outdoor scenes mostly stays the same as the camera usually sees the sky; (ii) the indoor sequences contain much more rotational motions, which cause difficulties for the pose network, while the motions of outdoor sequences are pre-dominantly translational, especially for driving datasets such as KITTI. In this paper, special considerations are given to those challenges and a set of good practices are consolidated for improving the performance of self-supervised monocular depth estimation in indoor environments. The proposed method mainly consists of two novel modules, i.e., a depth factorization module and a residual pose estimation module, each of which is designed to respectively tackle the aforementioned challenges. The effectiveness of each module is shown through a carefully conducted ablation study and the demonstration of the state-of-the-art performance on three indoor datasets, i.e., EuRoC, NYUv2 and 7-Scenes.

Recent advances in Convolutional Neural Network (CNN) model interpretability have led to impressive progress in visualizing and understanding model predictions. In particular, gradient-based visual attention methods have driven much recent effort in using visual attention maps as a means for visual explanations. A key problem, however, is these methods are designed for classification and categorization tasks, and their extension to explaining generative models, e.g., variational autoencoders (VAE) is not trivial. In this work, we take a step towards bridging this crucial gap, proposing the first technique to visually explain VAEs by means of gradient-based attention. We present methods to generate visual attention from the learned latent space, and also demonstrate such attention explanations serve more than just explaining VAE predictions. We show how these attention maps can be used to localize anomalies in images, demonstrating state-of-the-art performance on the MVTec-AD dataset. We also show how they can be infused into model training, helping bootstrap the VAE into learning improved latent space disentanglement, demonstrated on the Dsprites dataset.

This paper addresses the challenging problem of generating videos with human action semantics. Unlike previous work which predict future frames in a single forward pass, this paper introduces the cycle constraints in both forward and backward passes in the generation of human actions. This is achieved by enforcing the appearance and motion consistency across a sequence of frames generated in the future. The approach consists of two stages. In the first stage, the pose of a human body is generated. In the second stage, an image generator is used to generate future frames by using (a) generated human poses in the future from the first stage, (b) the single observed human pose, and (c) the single corresponding future frame. The experiments are performed on three datasets: Weizmann dataset involving simple human actions, Penn Action dataset and UCF-101 dataset containing complicated human actions, especially in sports. The results from these experiments demonstrate the effectiveness of the proposed approach.