A Better Baseline for AVA
arXiv:1807.10066
Abstract
We introduce a simple baseline for action localization on the AVA dataset. The model builds upon the Faster R-CNN bounding box detection framework, adapted to operate on pure spatiotemporal features - in our case produced exclusively by an I3D model pretrained on Kinetics. This model obtains 21.9% average AP on the validation set of AVA v2.1, up from 14.5% for the best RGB spatiotemporal model used in the original AVA paper (which was pretrained on Kinetics and ImageNet), and up from 11.3 of the publicly available baseline using a ResNet101 image feature extractor, that was pretrained on ImageNet. Our final model obtains 22.8%/21.9% mAP on the val/test sets and outperforms all submissions to the AVA challenge at CVPR 2018.
ActivityNet Workshop (AVA Challenge), CVPR 2018
References in corpus (5)
Cited by in corpus (16)
- SlowFast Networks for Video Recognition
- ROAD: The ROad event Awareness Dataset for Autonomous Driving
- Long-Term Feature Banks for Detailed Video Understanding
- Video action detection by learning graph-based spatio-temporal interactions
- Learning Discriminative Motion Features Through Detection
- Action Genome: Actions as Composition of Spatio-temporal Scene Graphs
- Actor-Context-Actor Relation Network for Spatio-Temporal Action Localization
- Group-Skeleton-Based Human Action Recognition in Complex Events
- Asynchronous Interaction Aggregation for Action Detection
- Three Branches: Detecting Actions With Richer Features
- Dance with Flow: Two-in-One Stream Action Detection
- Spatio-temporal Video Re-localization by Warp LSTM
- Towards Streaming Perception
- Attend What You Need: Motion-Appearance Synergistic Networks for Video Question Answering
- Spatio-Temporal Context for Action Detection
- Hierarchical Graph-RNNs for Action Detection of Multiple Activities