Delta Sampling R-BERT for limited data and low-light action recognition
It addresses action recognition in the dark, a domain-specific problem, with incremental improvements.
The paper tackles action recognition in low-light conditions using a smaller dataset (ARID), achieving an 11% improvement over previous baseline models.
We present an approach to perform supervised action recognition in the dark. In this work, we present our results on the ARID dataset. Most previous works only evaluate performance on large, well illuminated datasets like Kinetics and HMDB51. We demonstrate that our work is able to achieve a very low error rate while being trained on a much smaller dataset of dark videos. We also explore a variety of training and inference strategies including domain transfer methodologies and also propose a simple but useful frame selection strategy. Our empirical results demonstrate that we beat previously published baseline models by 11%.