Trajectory-Pooled Deep-Convolutional Descriptors (TDD)

Here we provide the code for the extraction of Trajectory-Pooled Deep-Convolutional Descriptors (TDD), from the following paper:

Action Recognition with Trajectory-Pooled Deep-Convolutional Descriptors
Limin Wang, Yu Qiao, and Xiaou Tang, in CVPR, 2015

Two-stream CNN models trained on the UCF101 dataset

First, we provide our trained two-stream CNN models on the split1 of UCF101 dataset, which achieve the recognition accuracy of 84.7%

"Spatial net model"

"Spatial net prototxt" "Temporal net model" "Temporal net prototxt"

TDD demo code

Here, a matlab demo code for TDD extraction is released.

• Step 1: Improved Trajectory Extraction

You need download our modified iDT feature code and compile it by yourself. Improved Trajectories

• Step 2: TVL1 Optical Flow Extraction

You need download our dense flow code and compile it by yourself. Dense Flow

• Step 3: Mat Caffe

You need download the public caffe toolbox. Our TDD code is compatatible with previous version of Caffe

• Step 4: TDD Extraction

Now you can run the matlab file "script_demo.m" to extract TDD features.

Questions

Contact - Limin Wang