Home

Awesome

BSN++: Complementary Boundary Regressor with Scale-Balanced Relation Modeling for Temporal Action Proposal Generation

A pytorch-version implementation codes of paper: "BSN++: Complementary Boundary Regressor with Scale-Balanced Relation Modeling for Temporal Action Proposal Generation", which is accepted in AAAI 2021.

[Arxiv Preprint]

Result

ANRecall
AR@133.7%
AR@547.8%
AR@1055.0%
AR@10075.3%
AUC66.74

Prerequisites

These code is implemented in Pytorch 1.5.1 + Python3 .

Download Datasets

The author rescaled the feature length of all videos to same length 100, and he provided the rescaled feature at BaiduCloud [Code:efy8].

Training and Testing of BSN++

All configurations of BSN++ are saved in opts.py, where you can modify training and model parameter.

  1. To train the BSN++:
python main.py --mode train
  1. To get the inference proposal of the validation videos and evaluate the proposals with recall and AUC:
python main.py --mode inference

Of course, you can complete all the process above in one line:

sh bsnpp.sh

Reference

This implementation largely borrows from BMN by JJBOY.

code:BMN

paper:BSN++: Complementary Boundary Regressor with Scale-Balanced Relation Modeling for Temporal Action Proposal Generation