No Arabic abstract
We introduce a novel Recurrent Neural Network-based algorithm for future video feature generation and action anticipation called feature mapping RNN. Our novel RNN architecture builds upon three effective principles of machine learning, namely parameter sharing, Radial Basis Function kernels and adversarial training. Using only some of the earliest frames of a video, the feature mapping RNN is able to generate future features with a fraction of the parameters needed in traditional RNN. By feeding these future features into a simple multi-layer perceptron facilitated with an RBF kernel layer, we are able to accurately predict the action in the video. In our experiments, we obtain 18% improvement on JHMDB-21 dataset, 6% on UCF101-24 and 13% improvement on UT-Interaction datasets over prior state-of-the-art for action anticipation.
One of the major challenges for autonomous vehicles in urban environments is to understand and predict other road users actions, in particular, pedestrians at the point of crossing. The common approach to solving this problem is to use the motion history of the agents to predict their future trajectories. However, pedestrians exhibit highly variable actions most of which cannot be understood without visual observation of the pedestrians themselves and their surroundings. To this end, we propose a solution for the problem of pedestrian action anticipation at the point of crossing. Our approach uses a novel stacked RNN architecture in which information collected from various sources, both scene dynamics and visual features, is gradually fused into the network at different levels of processing. We show, via extensive empirical evaluations, that the proposed algorithm achieves a higher prediction accuracy compared to alternative recurrent network architectures. We conduct experiments to investigate the impact of the length of observation, time to event and types of features on the performance of the proposed method. Finally, we demonstrate how different data fusion strategies impact prediction accuracy.
We present an approach for weakly supervised learning of human actions. Given a set of videos and an ordered list of the occurring actions, the goal is to infer start and end frames of the related action classes within the video and to train the respective action classifiers without any need for hand labeled frame boundaries. To address this task, we propose a combination of a discriminative representation of subactions, modeled by a recurrent neural network, and a coarse probabilistic model to allow for a temporal alignment and inference over long sequences. While this system alone already generates good results, we show that the performance can be further improved by approximating the number of subactions to the characteristics of the different action classes. To this end, we adapt the number of subaction classes by iterating realignment and reestimation during training. The proposed system is evaluated on two benchmark datasets, the Breakfast and the Hollywood extended dataset, showing a competitive performance on various weak learning tasks such as temporal action segmentation and action alignment.
In this report, the technical details of our submission to the EPIC-Kitchens Action Anticipation Challenge 2021 are given. We developed a hierarchical attention model for action anticipation, which leverages Transformer-based attention mechanism to aggregate features across temporal dimension, modalities, symbiotic branches respectively. In terms of Mean Top-5 Recall of action, our submission with team name ICL-SJTU achieved 13.39% for overall testing set, 10.05% for unseen subsets and 11.88% for tailed subsets. Additionally, it is noteworthy that our submission ranked 1st in terms of verb class in all three (sub)sets.
Anticipating human motion depends on two factors: the past motion and the persons intention. While the first factor has been extensively utilized to forecast short sequences of human motion, the second one remains elusive. In this work we approximate a persons intention via a symbolic representation, for example fine-grained action labels such as walking or sitting down. Forecasting a symbolic representation is much easier than forecasting the full body pose with its complex inter-dependencies. However, knowing the future actions makes forecasting human motion easier. We exploit this connection by first anticipating symbolic labels and then generate human motion, conditioned on the human motion input sequence as well as on the forecast labels. This allows the model to anticipate motion changes many steps ahead and adapt the poses accordingly. We achieve state-of-the-art results on short-term as well as on long-term human motion forecasting.
Minimally invasive surgery mainly consists of a series of sub-tasks, which can be decomposed into basic gestures or contexts. As a prerequisite of autonomic operation, surgical gesture recognition can assist motion planning and decision-making, and build up context-aware knowledge to improve the surgical robot control quality. In this work, we aim to develop an effective surgical gesture recognition approach with an explainable feature extraction process. A Bidirectional Multi-Layer independently RNN (BML-indRNN) model is proposed in this paper, while spatial feature extraction is implemented via fine-tuning of a Deep Convolutional Neural Network(DCNN) model constructed based on the VGG architecture. To eliminate the black-box effects of DCNN, Gradient-weighted Class Activation Mapping (Grad-CAM) is employed. It can provide explainable results by showing the regions of the surgical images that have a strong relationship with the surgical gesture classification results. The proposed method was evaluated based on the suturing task with data obtained from the public available JIGSAWS database. Comparative studies were conducted to verify the proposed framework. Results indicated that the testing accuracy for the suturing task based on our proposed method is 87.13%, which outperforms most of the state-of-the-art algorithms.