출처 : Faster-R-CNN Install on Ubuntu 16.04(GTX1080 CUDA 8.0,cuDNN 5.1)

Caffe Faster R CNN Training

1. 개요

2015년에 Microsoft에서 Caffe를 포함하여 Faster-R-CNN 소스코드 배포

• 추가 적으로 caffe를 설치 할 필요 없음
• 버클리대 공식 caffe에는 RoI pooling이 구현 되어 있지 않음

테스트 환경 (ubuntu 16.04)

• 그래픽카드는 GTX 1080이며 CUDA 8.0과 cuDNN 5.1을 사용한다.

2. 설치 (Google Cloud , Docker)

docker pull adioshun/faster-rcnn:20170808r1

sudo nvidia-docker run -i -t -p 2222:2222 -p 8585:8585 --volume /home/hjlim99/docker:/root --name 'rcnn2' adioshun/faster-rcnn:latest /bin/bash

설치 확인

• Download pre-computed Faster R-CNN detectors : ./data/scripts/fetch_faster_rcnn_models.sh
• cd $FRCN_ROOT : ./tools/demo.py

fetch_faster_rcnn_models.sh 수정 : URL=https://dl.dropboxusercontent.com/s/o6ii098bu51d139/$FILE
/home/py-faster-rcnn/lib/fast_rcnn/config.py : set __C.USE_GPU_NMS = False

2. 학습

Good to Great

$ ./data/scripts/fetch_imagenet_models.sh

time ./tools/train_net.py --gpu ${GPU_ID} \
  --solver models/${PT_DIR}/${NET}/faster_rcnn_end2end/solver.prototxt \
  --weights data/imagenet_models/${NET}.v2.caffemodel \
  --imdb ${TRAIN_IMDB} \
  --iters ${ITERS} \
  --cfg experiments/cfgs/faster_rcnn_end2end.yml \
  ${EXTRA_ARGS}

3. 테스트

./tools/test_net.py \ 
--gpu 0 \
--def models/pascal_voc/ZF/faster_rcnn_end2end/test.prototxt \
--net ./data/faster_rcnn_models/ZF_faster_rcnn_final.caffemodel \
--imdb voc_2007_test \
--cfg experiments/cfgs/faster_rcnn_end2end.yml

4. FineTuning

Use Faster RCNN and ResNet codes for object detection and image classification with your own training data

입력 동영상 변경

/workspace/py-faster-rcnn/tools/demo.py

    plt.savefig('demo_results/'+image_name)
    plt.close('all')

def vis_detections(im, class_name, dets, image_name, thresh=0.5):
    """Draw detected bounding boxes."""
    inds = np.where(dets[:, -1] >= thresh)[0]
    #if len(inds) == 0:
    #    return

PATH_TO_TEST_IMAGES_DIR = ''
im_names = [ os.path.join(PATH_TO_TEST_IMAGES_DIR, 'frame{}.jpg'.format(i)) for i in range(1, 11561) ]

#im_names = ['000456.jpg', '000542.jpg', '001150.jpg',
#            '001763.jpg', '004545.jpg']
for im_name in im_names:
    print '~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~'
    print 'Demo for {}'.format(im_name)
    demo(net, im_name)

입력 이미지 저장 위치 :/workspace/py-faster-rcnn/data/demo 저장 폴더 미리 생성 : /workspace/py-faster-rcnn/tools/demo_results/