Recent methods for video action recognition have reached outstanding performances on existing benchmarks. However, they tend to leverage context such as scenes or objects instead of focusing on understanding the human action itself. For instance, a tennis field leads to the prediction playing tennis irrespectively of the actions performed in the video. In contrast, humans have a more complete understanding of actions and can recognize them without context. The best example of out-of-context actions are mimes, that people can typically recognize despite missing relevant objects and scenes. In this paper, we propose to benchmark action recognition methods in the absence of context. We therefore introduce a novel dataset, Mimetics, consisting of mimed actions for a subset of 50 classes from the Kinetics benchmark. Our experiments show that state-of-the-art 3D convolutional neural networks obtain disappointing results on such videos, highlighting the lack of true understanding of the human actions. Body language, captured by human pose and motion, is a meaningful cue to recognize out-of-context actions. We thus evaluate several pose-based baselines, either based on explicit 2D or 3D pose estimates, or on transferring pose features to the action recognition problem. This last method, less prone to inherent pose estimation noise, performs better than the other pose-based baselines, suggesting that an explicit pose representation might not be optimal for real-world action recognition.
Also available on arxiv.