np.random.seed(cfg.RNG_SEED)
caffe.set_random_seed(cfg.RNG_SEED)
# set up caffe
caffe.set_mode_gpu()
caffe.set_device(args.gpu_id)
output_dir = "./"
imdb = []
# 训练中并没有使用到imdb的信息，也不清楚那是个什么玩意，无伤大雅
# roidb中记录了文本实例，以及对应图片路径，之类的信息
print args.imdb_name
if args.imdb_name == "MSRA_TRAIN":
output_dir = "."
roidb = get_rroidb("train")
else:
imdb, roidb = combined_roidb(args.imdb_name)
print('{:d} roidb entries'.format(len(roidb)))
#print roidb
output_dir = get_output_dir(imdb)
print('{:d} roidb entries'.format(len(roidb)))
print('Output will be saved to `{:s}`'.format(output_dir))
train_net(args.solver,
roidb,
output_dir,
pretrained_model=args.pretrained_model,
max_iters=args.max_iters)
#encoding:utf-8
import numpy as np
import cv2
from rotation.data_extractor import get_rroidb
from rotation.r_roidb import add_rbbox_regression_targets
from rotation.r_minibatch import r_get_minibatch
from fast_rcnn.config import cfg, cfg_from_file, cfg_from_list, get_output_dir
from rotation.rt_train import filter_roidb
cfg_from_file('../../experiments/cfgs/faster_rcnn_end2end.yml')
for c in cfg:
print c, ':', cfg[c]
roidb = get_rroidb()
print 'len', len(roidb)
#print roidb[0]
roidb = filter_roidb(roidb)
print 'len_after', len(roidb)
print roidb[0]
bbox_means, bbox_stds = add_rbbox_regression_targets(roidb) #To test
#print roidb[0]
#print 'bbox_means',bbox_means
#print 'bbox_stds',bbox_stds
blobs = r_get_minibatch(roidb, 2)
im = blobs['data'].transpose(0, 2, 3, 1)[0]
gt_boxes = blobs['gt_boxes']
info = blobs['im_info']
#print 'im:',im
# Warmup on a dummy image
im = 128 * np.ones((300, 500, 3), dtype=np.uint8)
for i in xrange(2):
_, _ = r_im_detect(net, im)
# im_names = []
# gt_boxes = []
# test_dir = './data/dataset/ICDAR_2015/ch4_test_images'
# for img in os.listdir(test_dir):
# # ignore files created by jupyter
# if img[0] == '.':
# continue
# im_names.append(os.path.join(test_dir, img))
# gt_boxes.append([0, 0, 0, 0, 0])
roidb = get_rroidb(args.imdb_name)
# The detection results will save in cood_dir in txt
cood_dir = "./data/dataset/ICDAR_2015/results_txt"
# for im_idx in range(len(im_names)):
for im_idx in range(min(10, len(roidb))):
im_path = roidb[im_idx]['image']
im_name = im_path.split('/')[-1].split('.')[0]
gt_boxes = roidb[im_idx]['boxes']
ori_gt_boxes = roidb[im_idx]['ori_boxes']
print '~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~'
print 'detect for {}'.format(im_name)
# vis_image(im_names[im_idx], demo(net, im_names[im_idx], gt_boxes[im_idx], cood_dir))
vis_detections(im_path, demo(net,
im_path,
if not args.randomize:
# fix the random seeds (numpy and caffe) for reproducibility
np.random.seed(cfg.RNG_SEED)
caffe.set_random_seed(cfg.RNG_SEED)
# set up caffe
caffe.set_mode_gpu()
caffe.set_device(args.gpu_id)
output_dir = "./"
imdb = []
print args.imdb_name
if args.imdb_name == "MSRA_TRAIN":
output_dir = "."
roidb = get_rroidb("train")
else:
imdb, roidb = combined_roidb(args.imdb_name)
print ('{:d} roidb entries'.format(len(roidb)))
#print roidb
output_dir = get_output_dir(imdb)
print ('{:d} roidb entries'.format(len(roidb)))
#output_dir = get_output_dir(imdb)
print ('Output will be saved to `{:s}`'.format(output_dir))
train_net(args.solver, roidb, output_dir,
pretrained_model=args.pretrained_model,
max_iters=args.max_iters)
def _get_next_minibatch(self):
"""Return the blobs to be used for the next minibatch.
If cfg.TRAIN.USE_PREFETCH is True, then blobs will be computed in a
separate process and made available through self._blob_queue.
"""
#随机抽取一张图片
db_inds = self._get_next_minibatch_inds()
minibatch_db = [self._roidb[i] for i in db_inds]
return r_get_rotate_minibatch(minibatch_db, self._num_classes) # D
if __name__ == '__main__':
np.set_printoptions(suppress=True)
cfg_from_file('experiments/cfgs/faster_rcnn_end2end.yml')
roidb = get_rroidb('train')
#for i in range(len(roidb)):
#print roidb[i]
#add_rbbox_regression_targets(roidb)
a_roidb = roidb[1]
print __file__, 'a_roidb[gt]', a_roidb['boxes']
print a_roidb['gt_classes']
src = cv2.imread(a_roidb['image'])
img = src[:]
#cv2.imshow('src',img)
#vis_image(img,a_roidb['boxes'])
#blobs = r_get_minibatch([a_roidb], 2)
blobs = r_get_rotate_minibatch([a_roidb], 3)
# ['gt_boxes', 'data', 'im_info']
print 'blobs : ', blobs.keys()
self._cur = 0
def _get_next_minibatch_inds(self):
"""Return the roidb indices for the next minibatch."""
# TODO(rbg): remove duplicated code
if self._cur + cfg.TRAIN.IMS_PER_BATCH >= len(self._roidb):
self._shuffle_roidb_inds()
db_inds = self._perm[self._cur:self._cur + cfg.TRAIN.IMS_PER_BATCH]
self._cur += cfg.TRAIN.IMS_PER_BATCH
return db_inds
def run(self):
print 'BlobFetcher started'
while True:
db_inds = self._get_next_minibatch_inds()
minibatch_db = [self._roidb[i] for i in db_inds]
blobs = r_get_minibatch(minibatch_db, self._num_classes)
self._queue.put(blobs)
if __name__ == '__main__':
roidb = get_rroidb()
#for i in range(len(roidb)):
#print roidb[i]
add_rbbox_regression_targets(roidb)
print roidb
print r_get_minibatch(roidb, 2)
if not args.randomize:
# fix the random seeds (numpy and caffe) for reproducibility
np.random.seed(cfg.RNG_SEED)
caffe.set_random_seed(cfg.RNG_SEED)
# set up caffe
caffe.set_mode_gpu()
caffe.set_device(args.gpu_id)
output_dir = "./"
imdb = []
print args.imdb_name
if args.imdb_name == "MSRA_TRAIN":
output_dir = "."
roidb = get_rroidb() # "train"
else:
imdb, roidb = combined_roidb(args.imdb_name)
print ('{:d} roidb entries'.format(len(roidb)))
#print roidb
output_dir = get_output_dir(imdb)
print ('{:d} roidb entries'.format(len(roidb)))
#output_dir = get_output_dir(imdb)
print ('Output will be saved to `{:s}`'.format(output_dir))
train_net(args.solver, roidb, output_dir,
pretrained_model=args.pretrained_model,
max_iters=args.max_iters)