/WSDDN

Weakly Supervised Deep Detection Networks (CVPR 2016)

Primary LanguageMatlab

Weakly Supervised Deep Detection Networks (WSDDN)

Installation

  1. Download and install MatConvNet
  2. Install this module with the package manager of MatConvNet vl_contrib:
    vl_contrib('install', 'WSDDN') ;
    vl_contrib('setup', 'WSDDN') ;
  1. If you want to train a WSDDN model, wsddn_train will automatically download the items below:

    a. PASCAL VOC 2007 devkit and dataset under data folder

    b. Pre-computed edge-boxes for trainval and test splits:

    c. Pre-trained network from MatConvNet website

  2. You can also download the pre-trained WSDDN model (VGGF-EB-BoxSc-SpReg). Note that it gives slightly different performance reported than in the paper (34.4% mAP instead of 34.5% mAP)

Demo

After completing the installation and downloading the required files, you are ready for the demo

            cd scripts;
            opts.modelPath = '....' ;
            opts.imdbPath = '....' ;
            opts.gpu = .... ;
            wsddn_demo(opts) ;
                        

Test

            addpath scripts;
            opts.modelPath = '....' ;
            opts.imdbPath = '....' ;
            opts.gpu = .... ;
            opts.vis = true ; % visualize
            wsddn_test(opts) ;
                        

Train

Download an ImageNet pre-trained model from http://www.vlfeat.org/matconvnet/pretrained/

            addpath scripts;
            opts.modelPath = '....' ;
            opts.imdbPath = '....' ;
            opts.train.gpus = .... ;
            [net,info] = wsddn_train(opts) ;
                        

Citing WSDDN

If you find the code useful, please cite:

    @inproceedings{Bilen16,
      author     = "Bilen, H. and Vedaldi, A.",
      title      = "Weakly Supervised Deep Detection Networks",
      booktitle  = "Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition",
      year       = "2016"
    }

Acknowledgement

Many thanks to Sam Albanie for his help with contrib package manager and other nameless heros who diligently found my bugs.

License

The analysis work performed with the program(s) must be non-proprietary work. Licensee and its contract users must be or be affiliated with an academic facility. Licensee may additionally permit individuals who are students at such academic facility to access and use the program(s). Such students will be considered contract users of licensee. The program(s) may not be used for commercial competitive analysis (such as benchmarking) or for any commercial activity, including consulting.