deep_nets

1. Download and process FLIC dataset
   cd data/FLIC; python process.py
   

2. Train detector:
   THEANO_FLAGS=mode=FAST_RUN,device=gpu0,floatX=float32 python train_detector.py conf.xml
   

3. Test detector:
   Change the epoch_no value in conf.xml. While training, models after every epoch are written to disk. You can use any one of those.
   THEANO_FLAGS=mode=FAST_RUN,device=gpu0,floatX=float32 python test_detector.py conf.xml