Activity recognition in videos
Dataset: Social videos EGO4D
Paper: https://arxiv.org/abs/2110.07058 (Links to an external site.)
Human activity recognition in the wild is a challenging task due to the inner challenges in computer vision (viewpoint or lighting variations, occlusion, among others). In this project, you are asked to explore and implement activity recognition frameworks able to recognize human activities in images or videos.
Relevant papers:
Dimiccoli, Mariella, Alejandro Cartas, and Petia Radeva. "Activity recognition from visual lifelogs: State of the art and future challenges." Multimodal Behavior Analysis in the Wild (2019): 121-134.
Cartas, Alejandro, et al. "Seeing and hearing egocentric actions: How much can we learn?." Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops. 2019
Cartas, Alejandro, et al. "Recognizing activities of daily living from egocentric images." Iberian Conference on Pattern Recognition and Image Analysis. Springer, Cham, 2017