D2-Net: Weakly-Supervised Action Localization via Discriminative Embeddings and Denoised Activations
mAP@0.5: 35.9mAP@0.1:0.7: -mAP@0.1:0.5: 51.4
In this task, the training data consists of videos with a list of activities in them without any temporal boundary annotations. However, while testing, given a video, the algorithm should recognize the activities in the video and also provide the start and end time.