Tree-Structured Policy based Progressive Reinforcement Learning for Temporally Language Grounding in Video (AAAI2020)

This repository contains the pytorch codes and trained models described in the paper "Tree-Structured Policy based Progressive Reinforcement Learning for Temporally Language Grounding in Video" By Jie Wu, Guanbin Li, Si Liu, Liang Lin. Paper

Motivation

Framework

Requirements

• Python 2.7
• Pytorch 0.4.1
• matplotlib
• The code is for Charades-STA dataset.

Visual Features

Please download the features in Features1, and put it in the "Dataset/Charades" folder.

Training and Testing Data

Please download the TrainingData in TrainingData, and put it in the "Dataset/Charades/ref_info" folder. Please download the TestingData in TestingData, and put it in the "Dataset/Charades/ref_info" folder.

Pre-trained models

We provide the pre-trained model for Charades-STA dataset, which can get 24.73 on R@1, IoU0.7 and 45.30 on R@1, IoU0.5: Models

Train

python train.py

Validate

python val.py

Test from Pre-trained Model

python test.py