Interactive Deep Learning Method for Segmenting Moving Objects

The git repository contains the code for the paper:

"Interactive Deep Learning Method for Segmenting Moving Objects" by Yi Wang, Zhiming Luo, Pierre-Marc Jodoin.

Accept at Special Issue on Scene Background Modeling and Initialization of Pattern Recognition Letters.

Citing

If you use this code, please cite our paper.

Requirements: software

Before use this code, you need have the MatConvNet toolbox installed which can be download at MatConvnet's home page. and the version we used is matconvnet-1.0-beta9.

Data: CDNet dataset and SMI dataset

The training data used in the paper can be download at:

(1) CDNet dataset

(2) SBMI dataset

Evaluation

For the CDNet dataset, we upload the result to CDNet for evaluation.

And the SBMI dataset, the evalutaion metrics were computed by using the code provide by CDNet.