def interval_scores(fusion_param):
# fusion_param = get_fusion_param()
camera_delta_s = viz_data_for_market(fusion_param)
for camera_delta in camera_delta_s:
for delta_s in camera_delta:
delta_s.sort()
gap_cnt = 5
camera_pair_travel_probs = [[
list() for _ in range(len(camera_delta_s[0]))
] for _ in range(len(camera_delta_s))]
for i, camera_delta in enumerate(camera_delta_s):
for j, delta_s in enumerate(camera_delta):
gap_width = (delta_s[-1] - delta_s[0]) / float(gap_cnt)
for k in range(gap_cnt):
left_bound = delta_s[0] + gap_width * k
right_bound = delta_s[1] + gap_width * (k + 1)
total_cnt = sum(map(len, camera_delta))
sp_cnt = len(delta_s)
camera_pair_travel_probs[i][j].append({
'left':
left_bound,
'right':
right_bound,
'prob':
sp_cnt / float(total_cnt)
# 'prob': (binary_search(delta_s, right_bound) - binary_search(delta_s, left_bound)) / float(total_cnt)
})
pickle_save(fusion_param['interval_pickle_path'], camera_pair_travel_probs)
return camera_pair_travel_probs
def compute_trainset_deltas(train_imgs_path, pickle_path):
train_tracks, camera_cnt = load_train_img_infos(train_imgs_path)
mid = calculate_mid_frame(train_tracks)
head_train_tracks = filter(lambda x: x[2] < mid, train_tracks)
deltas = [[list() for _ in range(camera_cnt)] for _ in range(camera_cnt)]
for query_track in head_train_tracks:
for gallery_track in head_train_tracks:
if query_track[0] == gallery_track[
0] and query_track[1] != gallery_track[1]:
delta = query_track[2] - gallery_track[2]
if abs(delta) < 60000:
deltas[query_track[1] - 1][gallery_track[1] -
1].append(delta)
pickle_save('head_' + pickle_path, deltas)
tail_train_tracks = filter(lambda x: x[2] > mid, train_tracks)
deltas = [[list() for _ in range(camera_cnt)] for _ in range(camera_cnt)]
for query_track in tail_train_tracks:
for gallery_track in tail_train_tracks:
if query_track[0] == gallery_track[
0] and query_track[1] != gallery_track[1]:
delta = query_track[2] - gallery_track[2]
if abs(delta) < 60000:
deltas[query_track[1] - 1][gallery_track[1] -
1].append(delta)
pickle_save('tail_' + pickle_path, deltas)
def save_market_test_truth():
ctrl_msg['data_folder_path'] = 'market_market-test'
fusion_param = get_fusion_param()
answer_path = fusion_param['answer_path']
answer_lines = read_lines(answer_path)
query_tracks = list()
for answer in answer_lines:
info = answer.split('_')
if 'bmp' in info[2]:
info[2] = info[2].split('.')[0]
if len(info) > 4 and 'jpe' in info[6]:
query_tracks.append([info[0], int(info[1][0]), int(info[2])])
else:
query_tracks.append(
[info[0],
int(info[1][1]),
int(info[2]),
int(info[1][3])])
gallery_path = fusion_param['gallery_path']
gallery_lines = read_lines(gallery_path)
gallery_tracks = list()
for gallery in gallery_lines:
info = gallery.split('_')
if 'bmp' in info[2]:
info[2] = info[2].split('.')[0]
if len(info) > 4 and 'jpe' in info[6]:
gallery_tracks.append([info[0], int(info[1][0]), int(info[2])])
else:
gallery_tracks.append(
[info[0],
int(info[1][1]),
int(info[2]),
int(info[1][3])])
gallery_tracks.extend(query_tracks)
print(len(gallery_tracks))
deltas = [[list() for j in range(6)] for i in range(6)]
for i, market_probe_track in enumerate(gallery_tracks):
if gallery_tracks[i][0] == 0 or gallery_tracks[i][0] == -1:
continue
for j in range(len(gallery_tracks)):
if gallery_tracks[i][0] == gallery_tracks[j][0] \
and i != j \
and gallery_tracks[i][3] == gallery_tracks[j][3] \
and gallery_tracks[i][1] != gallery_tracks[j][1]:
if gallery_tracks[i][1] == 4 and gallery_tracks[j][1] - 1 == 5:
if j >= 19732:
print gallery_tracks[i][2] - gallery_tracks[j][2]
deltas[gallery_tracks[i][1] -
1][gallery_tracks[j][1] -
1].append(gallery_tracks[i][2] -
gallery_tracks[j][2])
for camera_delta in deltas:
for delta_s in camera_delta:
delta_s.sort()
pickle_save('true_market_pg.pck', deltas)
def compute_trainset_deltas(train_imgs_path, pickle_path):
train_tracks, camera_cnt = load_train_img_infos(train_imgs_path)
deltas = [[list() for _ in range(camera_cnt)] for _ in range(camera_cnt)]
for query_track in train_tracks:
for gallery_track in train_tracks:
if query_track[0] == gallery_track[0]:
delta = query_track[2] - gallery_track[2]
if abs(delta) < 300000:
deltas[query_track[1]-1][gallery_track[1] - 1].append(delta)
pickle_save(pickle_path, deltas)
def store_sorted_deltas(fusion_param):
# 时空模型构建核心函数，
# 存储每对摄像头的时间差分布，
# 共6×6个数组，每个数组长度为该对摄像头统计到的时间差数目
# 这个函数运行前会删除distribution_pickle_path： sorted_deltas.pickle，因此也不会有缓存问题
camera_delta_s = viz_data_for_market(fusion_param)
# 对时间差做排序，在预测的时候能快速定位时间差位置，得到时间差的概率
for camera_delta in camera_delta_s:
for delta_s in camera_delta:
delta_s.sort()
# for python
safe_remove(fusion_param['distribution_pickle_path'])
pickle_save(fusion_param['distribution_pickle_path'], camera_delta_s)
def save_grid_train_truth():
ctrl_msg['data_folder_path'] = 'market_grid-cv0-train'
fusion_param = get_fusion_param()
market_train_tracks = train_tracks(fusion_param)
deltas = [[list() for j in range(6)] for i in range(6)]
for i, market_train_track in enumerate(market_train_tracks):
for j in range(0, len(market_train_tracks)):
if market_train_tracks[i][0] == market_train_tracks[j][0] \
and i != j \
and market_train_tracks[i][1] != market_train_tracks[j][1]:
deltas[market_train_tracks[i][1] -
1][market_train_tracks[j][1] -
1].append(market_train_tracks[i][2] -
market_train_tracks[j][2])
for camera_delta in deltas:
for delta_s in camera_delta:
delta_s.sort()
pickle_save('true_grid-cv0_train.pck', deltas)
def split_trainset_deltas(train_imgs_path, pickle_path, slice_cnt):
train_tracks, camera_cnt = load_train_img_infos(train_imgs_path)
slice_bounds = calculate_part_frame(train_tracks, slice_cnt)
split_tracks = list()
for i in range(slice_cnt / 2 + 1):
split_tracks.append(
filter(
lambda x: slice_bounds[i] < x[2] < slice_bounds[
i + slice_cnt / 2], train_tracks))
for i, split_track in enumerate(split_tracks):
deltas = [[list() for _ in range(camera_cnt)]
for _ in range(camera_cnt)]
for query_track in split_track:
for gallery_track in split_track:
if query_track[0] == gallery_track[
0] and query_track[1] != gallery_track[1]:
delta = query_track[2] - gallery_track[2]
if abs(delta) < 60000:
deltas[query_track[1] - 1][gallery_track[1] -
1].append(delta)
pickle_save(('part%d_' % i) + pickle_path, deltas)
def get_predict_delta_tracks(fusion_param, useful_predict_limit=10, random=False, diff_person=False, use_real_st=False):
# 获取左图列表
answer_path = fusion_param['answer_path']
answer_lines = read_lines(answer_path)
camera_cnt = 6
real_tracks = list()
for answer in answer_lines:
info = answer.split('_')
if 'bmp' in info[2]:
#
info[2] = info[2].split('.')[0]
if len(info) > 4 and 'jpe' in info[6]:
# grid
real_tracks.append([info[0], int(info[1][0]), int(info[2]), 1])
elif 'f' in info[2]:
real_tracks.append([info[0], int(info[1][1]), int(info[2][1:-5]), 1])
camera_cnt = 8
else:
# market
real_tracks.append([info[0], int(info[1][1]), int(info[2]), int(info[1][3])])
print 'left image ready'
# 获取右图列表
renew_pid_path = fusion_param['renew_pid_path']
predict_lines = read_lines(renew_pid_path)
print 'predict images ready'
# 左图中的人在右图可能出现在6个摄像头中
camera_delta_s = [[list() for j in range(camera_cnt)] for i in range(camera_cnt)]
person_cnt = len(answer_lines)
# market1501数据集有六个序列，只有同一个序列才能计算delta
if random:
useful_predict_limit = max(len(predict_lines)/100, 10)
for i, line in enumerate(predict_lines):
predict_pids = line.split(' ')
useful_cnt = 0
for j, predict_pid in enumerate(predict_pids):
if useful_cnt > useful_predict_limit:
break
if random:
predict_pid = randint(0, person_cnt - 1)
elif diff_person:
predict_pid = randint(10, person_cnt - 1)
else:
# todo transfer: if predict by python, start from 0, needn't minus 1
predict_pid = int(predict_pid)
predict_pid = int(predict_pid)
# same seq
# todo ignore same camera track
if real_tracks[i][3] == real_tracks[predict_pid][3] and real_tracks[i][1] != real_tracks[predict_pid][1]:
# and pid equal: real st
# if use_real_st and random or real_tracks[i][0] == real_tracks[predict_pid][0]:
if True:
useful_cnt += 1
delta = real_tracks[i][2] - real_tracks[predict_pid][2]
if abs(delta) < 1000000:
camera_delta_s[real_tracks[i][1] - 1][real_tracks[predict_pid][1] - 1].append(delta)
print 'deltas collected'
for camera_delta in camera_delta_s:
for delta_s in camera_delta:
delta_s.sort()
print 'deltas sorted'
# for python
safe_remove(fusion_param['distribution_pickle_path'])
pickle_save(fusion_param['distribution_pickle_path'], camera_delta_s)
print 'deltas saved'
return camera_delta_s
print('kmeans:')
print(self.kmeans_score / self.tracklet_cnt)
print('ap:')
print(self.spectral_score / self.tracklet_cnt)
def arg_parse():
parser = argparse.ArgumentParser(description='eval on txt')
parser.add_argument('--train_list',
default='../data/duke/train.list',
type=str,
help='')
parser.add_argument(
'--transfer_feature',
default=
'/home/cwh/coding/taudl_pyt/baseline/eval/grid_duke-train/train_ft.mat',
type=str,
help='')
parser.add_argument('--transfer', default='grid_duke', type=str, help='')
opt = parser.parse_args()
return opt
if __name__ == '__main__':
opt = arg_parse()
# a = pickle_load(opt.transfer+'_cluster.pck')
pseudo_camera_tracks = single_camera_time_cluster(opt.train_list)
c = TrackFeatureCluster(pseudo_camera_tracks, opt.transfer_feature) #0.31
c.fit()
pickle_save(opt.transfer + '_cluster.pck', c.cameras_tracks)