def predict_eval(target, predict_path):
if target == 'market' or target == 'market-r':
market_result_eval(predict_path)
elif target == 'markets1' or target == 'markets1-r':
market_result_eval(predict_path, TEST='/home/cwh/coding/markets1/test', QUERY='/home/cwh/coding/markets1/probe')
elif 'grid' in target:
grid_result_eval(predict_path)
def predict_eval(target, predict_path):
if target == 'market' or target == 'market-r':
market_result_eval(predict_path,
TEST = '/home/cwh/coding/Market-1501/test', QUERY = '/home/cwh/coding/Market-1501/probe')
elif 'grid' in target:
grid_result_eval(predict_path)
elif 'duke' in target:
market_result_eval(predict_path, log_path='duke_eval.log', TEST = '/home/cwh/coding/DukeMTMC-reID/test', QUERY = '/home/cwh/coding/DukeMTMC-reID/probe')
net = Model(inputs=[net.get_layer('resnet50').get_input_at(0)],
outputs=[net.get_layer('resnet50').get_output_at(0)])
# net = Model(inputs=[net.input], outputs=[net.get_layer('avg_pool').output])
test_f, test_info = extract_feature(gallery_path, net)
query_f, query_info = extract_feature(probe_path, net)
result, result_argsort = sort_similarity(query_f, test_f)
for i in range(len(result)):
result[i] = result[i][result_argsort[i]]
result = np.array(result)
safe_remove(pid_path)
safe_remove(score_path)
np.savetxt(pid_path, result_argsort, fmt='%d')
np.savetxt(score_path, result, fmt='%.4f')
if __name__ == '__main__':
source = 'market'
target = 'market'
for i in range(10):
net = load_model('../pretrain/' + source +
'_multi_sd_pretrain_%d.h5' % i)
target_path = '/home/cwh/coding/Market-1501'
probe_path = target_path + '/probe'
gallery_path = target_path + '/test'
train_path = target_path + '/train'
pid_path = 'ret_train_pid.txt'
score_path = 'ret_train_score.txt'
# train_predict(net, train_path, pid_path, score_path)
test_predict(net, probe_path, gallery_path, pid_path, score_path)
market_result_eval(pid_path, 'market_eval.txt', gallery_path,
probe_path)
def predict_eval(target, predict_path):
if target == 'market' or target == 'market-r':
market_result_eval(predict_path,
TEST='/home/cwh/coding/Market-1501/test',
QUERY='/home/cwh/coding/Market-1501/probe')
elif 'grid' in target:
grid_result_eval(predict_path)
elif 'duke' == target:
market_result_eval(predict_path,
log_path='duke_eval.log',
TEST='/home/cwh/coding/DukeMTMC-reID/test',
QUERY='/home/cwh/coding/DukeMTMC-reID/probe')
elif 'dukehead' == target:
market_result_eval(predict_path,
log_path='duke_head_eval.log',
TEST='/home/cwh/coding/dukehead/test',
QUERY='/home/cwh/coding/dukehead/probe')
elif 'duketail' == target:
market_result_eval(predict_path,
log_path='duke_tail_eval.log',
TEST='/home/cwh/coding/duketail/test',
QUERY='/home/cwh/coding/duketail/probe')
elif 'duketqtail' == target:
market_result_eval(predict_path,
log_path='duke_tqtail_eval.log',
TEST='/home/cwh/coding/duketqtail/test',
QUERY='/home/cwh/coding/duketqtail/probe')
elif 'dukequerytail' == target:
market_result_eval(predict_path,
log_path='duke_querytail_eval.log',
TEST='/home/cwh/coding/dukequerytail/test',
QUERY='/home/cwh/coding/dukequerytail/probe')
def grid_eval(source, transform_dir):
target = 'grid'
for i in range(10):
test_pair_predict(source + '_pair_pretrain.h5',
transform_dir + 'cross%d' % i + '/probe',
transform_dir + 'cross%d' % i + '/test',
source + '_' + target + '_pid.log',
source + '_' + target + '_score.log')
grid_result_eval(source + '_' + target + '_pid.log', 'gan.log')
def market_eval(source, transform_dir):
target = 'market'
test_pair_predict(source + '_model.h5', transform_dir + '/query',
transform_dir + '/bounding_box_test',
source + '_' + target + '_pid.log',
source + '_' + target + '_score.log')
if __name__ == '__main__':
os.environ["CUDA_VISIBLE_DEVICES"] = "1"
market_eval('market', '../dataset/Market-1501-v15.09.15')
market_result_eval(
'market_market_pid.log',
TEST='../dataset/Market-1501-v15.09.15/bounding_box_test',
QUERY='../dataset/Market-1501-v15.09.15/query')
# grid_eval('market', '/home/person/grid_train_probe_gallery/cross0')
#grid_result_eval('/home/person/TrackViz/data/market_grid-cv0-test/cross_filter_pid.log')
pair_generator_by_rank_list(train_images, batch_size, similar_persons, similar_matrix, train=False),
target,
batch_size=batch_size, num_classes=class_count
)
if __name__ == '__main__':
# sources = ['cuhk_grid_viper_mix']
sources = ['cuhk']
target = 'market'
pair_model('../pretrain/cuhk_pair_pretrain.h5', 751)
for source in sources:
pair_pretrain_on_dataset(source, target)
transform_dir = '/home/cwh/coding/Market-1501'
safe_remove('pair_transfer_pid.log')
test_pair_predict('market_pair_pretrain.h5',
transform_dir + '/probe', transform_dir + '/test',
'pair_transfer_pid.log', 'pair_transfer_score.log')
market_result_eval('pair_transfer_pid.log', TEST='/home/cwh/coding/Market-1501/test',
QUERY='/home/cwh/coding/Market-1501/probe')
# sources = ['grid-cv-%d' % i for i in range(10)]
# for source in sources:
# softmax_pretrain_on_dataset(source,
# project_path='/home/cwh/coding/rank-reid',
# dataset_parent='/home/cwh/coding')
# pair_pretrain_on_dataset(source,
# project_path='/home/cwh/coding/rank-reid',
# dataset_parent='/home/cwh/coding')
if __name__ == '__main__':
# train_pair_predict('market_pair_pretrain.h5', '/home/cwh/coding/grid_train_probe_gallery/cross0/train', 'market_grid_pid.txt', 'market_grid_score.txt')
# grid_test_base_eval('../baseline/0.ckpt', 'grid_cross0_single')
# cross0: [0.072, 0.144, 0.184, 0.232, 0.408]
# cross0_gan: [0.136, 0.344, 0.416, 0.544, 0.648]
# grid_test_pair_eval('../transfer/pair_pretrain.h5', 'grid_cross0_pair_pretrain')
# [0.088, 0.16, 0.2, 0.296, 0.456]
# grid_test_pair_eval('../transfer/pair_transfer.h5', 'grid_cross0_transfer')
# [0.192, 0.312, 0.376, 0.496, 0.624] epoch7
# [0.192, 0.312, 0.392, 0.48, 0.648] st epoch3
# grid_test_rank_eval('../transfer/rank_transfer.h5', 'grid_cross0_rank_transfer')
# [0.184, 0.304, 0.344, 0.456, 0.656]
# test_pair_predict('grid_softmax_pretrain.h5', '/home/cwh/coding/Market-1501/probe', '/home/cwh/coding/Market-1501/test', 'market_market_pid_test.txt', 'market_market_score_test.txt')
# test_pair_predict('market_softmax_pretrain.h5', '/home/wxt/reid_data-gan/cross0_gan/probe', '/home/wxt/reid_data-gan/cross0_gan/test', 'market_market_pid_test_gan.txt', 'market_market_score_test_gan.txt')
# test_pair_predict('market_pair_pretrain.h5', '/home/cwh/coding/grid_train_probe_gallery/cross0/probe', '/home/cwh/coding/grid_train_probe_gallery/cross0/test', 'market_market_pid_test_softmax.txt', 'market_market_score_test_softmax.txt')
# grid_result_eval('market_market_pid_test_softmax.txt', './gan.log')
# [0.104, 0.176, 0.264, 0.312, 0.416]
# grid_eval('market', '/home/wxt/ReidGAN/transformgrid2marketstyle')
# market_eval('grid', '/home/wxt/ReidGAN/market2grid_style')
market_eval('market', '/home/cwh/coding/Market-1501')
market_result_eval('market_market_pid.log')
# market_result_eval('/home/cwh/coding/TrackViz/data/market_market-test/cross_filter_pid.log')
# test_rank_predict('../transfer/rank_transfer_test.h5',
# '/home/cwh/coding/Market-1501/probe', '/home/cwh/coding/Market-1501/test',
# 'rank_pid.log', 'rank_ac.log')
# market_result_eval('rank_pid.log')
# grid_result_eval('/home/cwh/coding/TrackViz/data/market_grid-cv0-test/cross_filter_pid.log')
# market_result_eval('/home/cwh/coding/TrackViz/data/market_market-test/cross_filter_pid.log')
