TASK	DATASET	MODEL	METRIC NAME	METRIC VALUE	GLOBAL RANK	REMOVE
Action Spotting	SoccerNet	CALF (Cioppa et al.)	Average-mAP	62.5	# 3
Action Spotting	SoccerNet-v2	CALF (Cioppa et al.)	Average-mAP	40.7	# 8

Badge	Markdown
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/a-context-aware-loss-function-for-action/action-spotting-on-soccernet)](https://paperswithcode.com/sota/action-spotting-on-soccernet?p=a-context-aware-loss-function-for-action)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/a-context-aware-loss-function-for-action/action-spotting-on-soccernet-v2)](https://paperswithcode.com/sota/action-spotting-on-soccernet-v2?p=a-context-aware-loss-function-for-action)`

A Context-Aware Loss Function for Action Spotting in Soccer Videos

CVPR 2020 · Anthony Cioppa, Adrien Deliège, Silvio Giancola, Bernard Ghanem, Marc Van Droogenbroeck, Rikke Gade, Thomas B. Moeslund ·

In video understanding, action spotting consists in temporally localizing human-induced events annotated with single timestamps. In this paper, we propose a novel loss function that specifically considers the temporal context naturally present around each action, rather than focusing on the single annotated frame to spot. We benchmark our loss on a large dataset of soccer videos, SoccerNet, and achieve an improvement of 12.8% over the baseline. We show the generalization capability of our loss for generic activity proposals and detection on ActivityNet, by spotting the beginning and the end of each activity. Furthermore, we provide an extended ablation study and display challenging cases for action spotting in soccer videos. Finally, we qualitatively illustrate how our loss induces a precise temporal understanding of actions and show how such semantic knowledge can be used for automatic highlights generation.