def _evaluate(self,
epoch,
dataset_name='',
queryloader=None,
galleryloader=None,
dist_metric='euclidean',
visrank=False,
visrank_topk=20,
save_dir='',
use_metric_cuhk03=False,
ranks=[1, 5, 10, 20]):
batch_time = AverageMeter()
self.model.eval()
print('Extracting features from query set ...')
qf, q_pids, q_camids = [], [], []
for batch_idx, data in enumerate(queryloader):
imgs, pids, camids = self._parse_data_for_eval(data)
if self.use_gpu:
imgs = imgs.cuda()
end = time.time()
features = self._extract_features(imgs)
batch_time.update(time.time() - end)
features = features.data.cpu()
qf.append(features)
q_pids.extend(pids)
q_camids.extend(camids)
qf = torch.cat(qf, 0)
q_pids = np.asarray(q_pids)
q_camids = np.asarray(q_camids)
print('Done, obtained {}-by-{} matrix'.format(qf.size(0), qf.size(1)))
print('Extracting features from gallery set ...')
gf, g_pids, g_camids = [], [], []
end = time.time()
for batch_idx, data in enumerate(galleryloader):
imgs, pids, camids = self._parse_data_for_eval(data)
if self.use_gpu:
imgs = imgs.cuda()
end = time.time()
features = self._extract_features(imgs)
batch_time.update(time.time() - end)
features = features.data.cpu()
gf.append(features)
g_pids.extend(pids)
g_camids.extend(camids)
gf = torch.cat(gf, 0)
g_pids = np.asarray(g_pids)
g_camids = np.asarray(g_camids)
print('Done, obtained {}-by-{} matrix'.format(gf.size(0), gf.size(1)))
print('Speed: {:.4f} sec/batch'.format(batch_time.avg))
distmat = metrics.compute_distance_matrix(qf, gf, dist_metric)
distmat = distmat.numpy()
print('Computing CMC and mAP ...')
cmc, mAP = metrics.evaluate_rank(distmat,
q_pids,
g_pids,
q_camids,
g_camids,
use_metric_cuhk03=use_metric_cuhk03)
print('** Results **')
print('mAP: {:.1%}'.format(mAP))
print('CMC curve')
for r in ranks:
print('Rank-{:<3}: {:.1%}'.format(r, cmc[r - 1]))
if visrank:
visualize_ranked_results(
distmat,
self.datamanager.return_testdataset_by_name(dataset_name),
save_dir=osp.join(save_dir, 'visrank-' + str(epoch + 1),
dataset_name),
topk=visrank_topk)
return cmc[0]
def _evaluate(self, epoch, dataset_name='', queryloader=None, galleryloader=None,
dist_metric='euclidean', normalize_feature=False, visrank=False,
visrank_topk=10, save_dir='', use_metric_cuhk03=False, ranks=(1, 5, 10, 20),
rerank=False, iteration=0):
batch_time = AverageMeter()
print('Extracting features from query set...')
qf, q_pids, q_camids = [], [], [] # query features, query person IDs and query camera IDs
for batch_idx, data in tqdm(enumerate(queryloader), 'Processing query...'):
imgs, pids, camids = self._parse_data_for_eval(data)
if self.use_gpu:
imgs = imgs.cuda()
end = time.time()
features = self._extract_features(imgs, data[3])
batch_time.update(time.time() - end)
features = features.data.cpu()
qf.append(features)
q_pids.extend(pids)
q_camids.extend(camids)
qf = torch.cat(qf, 0)
q_pids = np.asarray(q_pids)
q_camids = np.asarray(q_camids)
print('Done, obtained {}-by-{} matrix'.format(qf.size(0), qf.size(1)))
print('Extracting features from gallery set...')
gf, g_pids, g_camids = [], [], [] # gallery features, gallery person IDs and gallery camera IDs
for batch_idx, data in tqdm(enumerate(galleryloader), 'Processing gallery...'):
imgs, pids, camids = self._parse_data_for_eval(data)
if self.use_gpu:
imgs = imgs.cuda()
end = time.time()
features = self._extract_features(imgs, data[3])
batch_time.update(time.time() - end)
features = features.data.cpu()
gf.append(features)
g_pids.extend(pids)
g_camids.extend(camids)
gf = torch.cat(gf, 0)
g_pids = np.asarray(g_pids)
g_camids = np.asarray(g_camids)
print('Done, obtained {}-by-{} matrix'.format(gf.size(0), gf.size(1)))
print('Speed: {:.4f} sec/batch'.format(batch_time.avg))
if normalize_feature:
print('Normalizing features with L2 norm...')
qf = F.normalize(qf, p=2, dim=1)
gf = F.normalize(gf, p=2, dim=1)
print('Computing distance matrix with metric={}...'.format(dist_metric))
distmat = metrics.compute_distance_matrix(qf, gf, dist_metric)
distmat = distmat.numpy()
if rerank:
print('Applying person re-ranking ...')
distmat_qq = metrics.compute_distance_matrix(qf, qf, dist_metric)
distmat_gg = metrics.compute_distance_matrix(gf, gf, dist_metric)
distmat = re_ranking(distmat, distmat_qq, distmat_gg)
print('Computing CMC and mAP ...')
cmc, mAP = metrics.evaluate_rank(
distmat,
q_pids,
g_pids,
q_camids,
g_camids,
use_metric_cuhk03=use_metric_cuhk03
)
if self.writer is not None:
self.writer.add_scalar('Val/{}/mAP'.format(dataset_name), mAP, epoch + 1)
for r in ranks:
self.writer.add_scalar('Val/{}/Rank-{}'.format(dataset_name, r), cmc[r - 1], epoch + 1)
print('** Results **')
print('mAP: {:.2%}'.format(mAP))
print('CMC curve')
for r in ranks:
print('Rank-{:<3}: {:.2%}'.format(r, cmc[r-1]))
if visrank:
visualize_ranked_results(
distmat,
self.datamanager.return_testdataset_by_name(dataset_name),
self.datamanager.data_type,
width=self.datamanager.width,
height=self.datamanager.height,
save_dir=osp.join(save_dir, 'visrank_' + dataset_name),
topk=visrank_topk
)
return cmc[0]
def evaluate(model,
query_loader=None,
gallery_loader=None,
dist_metric='euclidean',
use_metric_cuhk03=False,
ranks=[1, 5, 10, 20],
rerank=False):
batch_time = avgmeter.AverageMeter()
def _feature_extraction(data_loader):
f_, pids_, camids_ = [], [], []
for batch_idx, data in enumerate(data_loader):
imgs, pids, camids = _parse_data_for_eval(data)
end = time.time()
_, features = model.predict_on_batch(imgs)
batch_time.update(time.time() - end)
f_.extend(features.numpy())
pids_.extend(pids.numpy())
camids_.extend(camids.numpy())
f_ = np.stack(f_)
pids_ = np.stack(pids_)
camids_ = np.stack(camids_)
return f_, pids_, camids_
print('Extracting features from query set ...')
qf, q_pids, q_camids = _feature_extraction(query_loader)
print('Done, obtained {}-by-{} matrix'.format(qf.shape[0], qf.shape[1]))
print('Extracting features from gallery set ...')
gf, g_pids, g_camids = _feature_extraction(gallery_loader)
print('Done, obtained {}-by-{} matrix'.format(gf.shape[0], gf.shape[1]))
print('Speed: {:.4f} sec/batch'.format(batch_time.avg))
# if normalize_feature:
# print('Normalzing features with L2 norm ...')
# qf = F.normalize(qf, p=2, dim=1)
# gf = F.normalize(gf, p=2, dim=1)
print(f'Computing distance matrix with metric={dist_metric} ...')
distmat = compute_distance_matrix(qf, gf, dist_metric)
if rerank:
print('Applying person re-ranking ...')
distmat_qq = compute_distance_matrix(qf, qf, dist_metric)
distmat_gg = compute_distance_matrix(gf, gf, dist_metric)
distmat = utils.re_ranking(distmat, distmat_qq, distmat_gg)
print('Computing CMC and mAP ...')
cmc, mAP = metrics.evaluate_rank(
distmat,
q_pids,
g_pids,
q_camids,
g_camids,
use_metric_cuhk03=use_metric_cuhk03
)
print('** Results **')
print('mAP: {:.1%}'.format(mAP))
print('CMC curve')
for r in ranks:
print('Rank-{:<3}: {:.1%}'.format(r, cmc[r - 1]))
# if visrank:
# visualize_ranked_results(
# distmat,
# self.datamanager.
# return_query_and_gallery_by_name(dataset_name),
# self.datamanager.data_type,
# width=self.datamanager.width,
# height=self.datamanager.height,
# save_dir=osp.join(save_dir, 'visrank_' + dataset_name),
# topk=visrank_topk
# )
return cmc[0]
def _evaluate(
self,
dataset_name='',
query_loader=None,
gallery_loader=None,
dist_metric='euclidean',
normalize_feature=False,
visrank=False,
visrank_topk=10,
save_dir='',
use_metric_cuhk03=False,
ranks=[1, 5, 10, 20],
rerank=False
):
batch_time = AverageMeter()
def _feature_extraction(data_loader):
f_, pids_, camids_ = [], [], []
for batch_idx, data in enumerate(data_loader):
imgs, pids, camids = self.parse_data_for_eval(data)
if self.use_gpu:
imgs = imgs.cuda()
end = time.time()
features = self.extract_features(imgs)
batch_time.update(time.time() - end)
features = features.cpu().clone()
f_.append(features)
pids_.extend(pids)
camids_.extend(camids)
f_ = torch.cat(f_, 0)
pids_ = np.asarray(pids_)
camids_ = np.asarray(camids_)
return f_, pids_, camids_
print('Extracting features from query set ...')
qf, q_pids, q_camids = _feature_extraction(query_loader)
print(qf.shape)
print('Done, obtained {}-by-{} matrix'.format(qf.size(0), qf.size(1)))
print('Extracting features from gallery set ...')
gf, g_pids, g_camids = _feature_extraction(gallery_loader)
print('Done, obtained {}-by-{} matrix'.format(gf.size(0), gf.size(1)))
print('Speed: {:.4f} sec/batch'.format(batch_time.avg))
if normalize_feature:
print('Normalzing features with L2 norm ...')
qf = F.normalize(qf, p=2, dim=1)
gf = F.normalize(gf, p=2, dim=1)
print(
'Computing distance matrix with metric={} ...'.format(dist_metric)
)
distmat = metrics.compute_distance_matrix(qf, gf, dist_metric)
distmat = distmat.numpy()
if rerank:
print('Applying person re-ranking ...')
distmat_qq = metrics.compute_distance_matrix(qf, qf, dist_metric)
distmat_gg = metrics.compute_distance_matrix(gf, gf, dist_metric)
distmat = re_ranking(distmat, distmat_qq, distmat_gg)
print('Computing CMC and mAP ...')
cmc, mAP = metrics.evaluate_rank(
distmat,
q_pids,
g_pids,
q_camids,
g_camids,
use_metric_cuhk03=use_metric_cuhk03
)
print('** Results **')
print('mAP: {:.1%}'.format(mAP))
print('CMC curve')
for r in ranks:
print('Rank-{:<3}: {:.1%}'.format(r, cmc[r - 1]))
if visrank:
visualize_ranked_results(
distmat,
self.datamanager.fetch_test_loaders(dataset_name),
self.datamanager.data_type,
width=self.datamanager.width,
height=self.datamanager.height,
save_dir=osp.join(save_dir, 'visrank_' + dataset_name),
topk=visrank_topk
)
return cmc[0], mAP
def evaluate(model,
queryloader,
galleryloader,
dist_metric='euclidean',
normalize_feature=False,
rerank=False,
return_distmat=False):
batch_time = AverageMeter()
model.eval()
with torch.no_grad():
print('Extracting features from query set ...')
qf, q_pids, q_camids = [], [], []
for batch_idx, (imgs, pids, camids, _) in enumerate(queryloader):
imgs = imgs.cuda()
end = time.time()
features = model(imgs)
batch_time.update(time.time() - end)
features = features.data.cpu()
qf.append(features)
q_pids.extend(pids)
q_camids.extend(camids)
qf = torch.cat(qf, 0)
q_pids = np.asarray(q_pids)
q_camids = np.asarray(q_camids)
print('Done, obtained {}-by-{} matrix'.format(qf.size(0), qf.size(1)))
print('Extracting features from gallery set ...')
gf, g_pids, g_camids = [], [], []
for batch_idx, (imgs, pids, camids, _) in enumerate(galleryloader):
imgs = imgs.cuda()
end = time.time()
features = model(imgs)
batch_time.update(time.time() - end)
features = features.data.cpu()
gf.append(features)
g_pids.extend(pids)
g_camids.extend(camids)
gf = torch.cat(gf, 0)
g_pids = np.asarray(g_pids)
g_camids = np.asarray(g_camids)
print('Done, obtained {}-by-{} matrix'.format(gf.size(0), gf.size(1)))
print('Speed: {:.4f} sec/batch'.format(batch_time.avg))
if normalize_feature:
print('Normalzing features with L2 norm ...')
qf = F.normalize(qf, p=2, dim=1)
gf = F.normalize(gf, p=2, dim=1)
print('Computing distance matrix with metric={} ...'.format(dist_metric))
distmat = compute_distance_matrix(qf, gf, dist_metric)
distmat = distmat.numpy()
if rerank:
print('Applying person re-ranking ...')
distmat_qq = compute_distance_matrix(qf, qf, dist_metric)
distmat_gg = compute_distance_matrix(gf, gf, dist_metric)
distmat = re_ranking(distmat, distmat_qq, distmat_gg)
print('Computing CMC and mAP ...')
cmc, mAP = evaluate_rank(distmat, q_pids, g_pids, q_camids, g_camids)
print('** Results **')
print('mAP: {:.1%}'.format(mAP))
print('CMC curve')
for r in [1, 5, 10, 20]:
print('Rank-{:<3}: {:.1%}'.format(r, cmc[r - 1]))
if return_distmat:
return distmat
return cmc[0]
def _evaluate_reid(self,
model,
epoch,
dataset_name='',
query_loader=None,
gallery_loader=None,
dist_metric='euclidean',
normalize_feature=False,
visrank=False,
visrank_topk=10,
save_dir='',
use_metric_cuhk03=False,
ranks=(1, 5, 10, 20),
rerank=False,
model_name='',
lr_finder=False):
def _feature_extraction(data_loader):
f_, pids_, camids_ = [], [], []
for _, data in enumerate(data_loader):
imgs, pids, camids = self.parse_data_for_eval(data)
if self.use_gpu:
imgs = imgs.cuda()
features = model(imgs),
features = features.data.cpu()
f_.append(features)
pids_.extend(pids)
camids_.extend(camids)
f_ = torch.cat(f_, 0)
pids_ = np.asarray(pids_)
camids_ = np.asarray(camids_)
return f_, pids_, camids_
qf, q_pids, q_camids = _feature_extraction(query_loader)
gf, g_pids, g_camids = _feature_extraction(gallery_loader)
if normalize_feature:
qf = F.normalize(qf, p=2, dim=1)
gf = F.normalize(gf, p=2, dim=1)
distmat = metrics.compute_distance_matrix(qf, gf, dist_metric)
distmat = distmat.numpy()
if rerank:
distmat_qq = metrics.compute_distance_matrix(qf, qf, dist_metric)
distmat_gg = metrics.compute_distance_matrix(gf, gf, dist_metric)
distmat = re_ranking(distmat, distmat_qq, distmat_gg)
cmc, mAP = metrics.evaluate_rank(distmat,
q_pids,
g_pids,
q_camids,
g_camids,
use_metric_cuhk03=use_metric_cuhk03)
if self.writer is not None and not lr_finder:
self.writer.add_scalar(
'Val/{}/{}/mAP'.format(dataset_name, model_name), mAP,
epoch + 1)
for r in ranks:
self.writer.add_scalar(
'Val/{}/{}/Rank-{}'.format(dataset_name, model_name, r),
cmc[r - 1], epoch + 1)
if not lr_finder:
print('** Results ({}) **'.format(model_name))
print('mAP: {:.2%}'.format(mAP))
print('CMC curve')
for r in ranks:
print('Rank-{:<3}: {:.2%}'.format(r, cmc[r - 1]))
if visrank and not lr_finder:
visualize_ranked_results(
distmat,
self.datamanager.fetch_test_loaders(dataset_name),
self.datamanager.data_type,
width=self.datamanager.width,
height=self.datamanager.height,
save_dir=osp.join(save_dir, 'visrank_' + dataset_name),
topk=visrank_topk)
return cmc[0]
def _evaluate(self, arch, epoch, dataset_name='', queryloader=None, galleryloader=None,
dist_metric='euclidean', normalize_feature=False, visrank=False,
visrank_topk=20, save_dir='', use_metric_cuhk03=False, ranks=[1, 5, 10, 20],
rerank=False, viscam=False, viscam_num=10, viscam_only=False):
with self.experiment.test():
if not viscam_only:
batch_time = AverageMeter()
combine_time = AverageMeter()
self.model.eval()
print('Extracting features from query set ...')
qf, q_pids, q_camids = [], [], [] # query features, query person IDs and query camera IDs
for batch_idx, data in enumerate(queryloader):
imgs, pids, camids = self._parse_data_for_eval(data)
if self.use_gpu:
imgs = imgs.cuda()
end = time.time()
features = self._extract_features(imgs)
batch_time.update(time.time() - end, len(pids), True)
features = features.data.cpu()
qf.append(features)
q_pids.extend(pids)
q_camids.extend(camids)
qf = torch.cat(qf, 0)
q_pids = np.asarray(q_pids)
q_camids = np.asarray(q_camids)
print('Done, obtained {}-by-{} matrix'.format(qf.size(0), qf.size(1)))
print('Extracting features from gallery set ...')
gf, g_pids, g_camids = [], [], [] # gallery features, gallery person IDs and gallery camera IDs
end = time.time()
for batch_idx, data in enumerate(galleryloader):
imgs, pids, camids = self._parse_data_for_eval(data)
if self.use_gpu:
imgs = imgs.cuda()
end = time.time()
features = self._extract_features(imgs)
batch_time.update(time.time() - end, len(pids), True)
features = features.data.cpu()
gf.append(features)
g_pids.extend(pids)
g_camids.extend(camids)
gf = torch.cat(gf, 0)
g_pids = np.asarray(g_pids)
g_camids = np.asarray(g_camids)
end = time.time()
num_images = len(g_pids)
self.combine_fn.train()
gf, g_pids = self.combine_fn(gf, g_pids, g_camids)
if self.save_embed:
assert osp.isdir(self.save_embed)
path = osp.realpath(self.save_embed)
np.save(path + '/gf-' + self.combine_method + '.npy', gf)
np.save(path + '/g_pids-' + self.combine_method + '.npy', g_pids)
combine_time.update(time.time() - end, num_images, True)
time.time() - end
gf = torch.tensor(gf, dtype=torch.float)
print('Done, obtained {}-by-{} matrix'.format(gf.size(0), gf.size(1)))
print('Speed: {:.4f} sec/image'.format(batch_time.avg + combine_time.avg))
if normalize_feature:
print('Normalzing features with L2 norm ...')
qf = F.normalize(qf, p=2, dim=1)
gf = F.normalize(gf, p=2, dim=1)
print('Computing distance matrix with metric={} ...'.format(dist_metric))
distmat = metrics.compute_distance_matrix(qf, gf, dist_metric)
distmat = distmat.numpy()
if rerank:
print('Applying person re-ranking ...')
distmat_qq = metrics.compute_distance_matrix(qf, qf, dist_metric)
distmat_gg = metrics.compute_distance_matrix(gf, gf, dist_metric)
distmat = re_ranking(distmat, distmat_qq, distmat_gg)
print('Computing CMC and mAP ...')
cmc, mAP = metrics.evaluate_rank(
distmat,
q_pids,
g_pids,
q_camids,
g_camids,
use_metric_cuhk03=use_metric_cuhk03
)
print('** Results **')
print('mAP: {:.1%}'.format(mAP))
print('CMC curve')
for r in ranks:
print('Rank-{:<3}: {:.1%}'.format(r, cmc[r-1]))
# write to Tensorboard and comet.ml
if not self.test_only:
rs = {'eval-rank-{:<3}'.format(r):cmc[r-1] for r in ranks}
self.writer.add_scalars('eval/ranks',rs,epoch)
self.experiment.log_metrics(rs,step=epoch)
self.writer.add_scalar('eval/mAP',mAP,epoch)
self.experiment.log_metric('eval-mAP',mAP,step=epoch)
print('Results written to tensorboard and comet.ml.')
if visrank:
visualize_ranked_results(
distmat,
self.datamanager.return_testdataset_by_name(dataset_name),
save_dir=osp.join(save_dir, 'visrank-'+str(epoch+1), dataset_name),
topk=visrank_topk
)
if viscam:
if arch == 'osnet_x1_0' or arch == 'osnet_custom':
# print(self.model)
visualize_cam(
model=self.model,
finalconv='conv5', # for OSNet
dataset=self.datamanager.return_testdataset_by_name(dataset_name),
save_dir=osp.join(save_dir, 'viscam-'+str(epoch+1), dataset_name),
num=viscam_num
)
elif arch == 'resnext50_32x4d':
# print(self.model)
visualize_cam(
model=self.model,
finalconv='layer4', # for resnext50
dataset=self.datamanager.return_testdataset_by_name(dataset_name),
save_dir=osp.join(save_dir, 'viscam-'+str(epoch+1), dataset_name),
num=viscam_num
)
if viscam_only:
raise RuntimeError('Stop exec because `viscam_only` is set to true.')
return cmc[0]
def test(model,
queryloader,
galleryloader,
use_gpu,
ranks=[1, 5, 10, 20],
return_distmat=False):
flip_eval = args.flip_eval
if flip_eval:
print('# Using Flip Eval')
batch_time = AverageMeter()
model.eval()
with torch.no_grad():
qf, q_pids, q_camids, q_paths = [], [], [], []
if flip_eval:
enumerator = enumerate(zip(queryloader[0], queryloader[1]))
else:
enumerator = enumerate(queryloader[0])
for batch_idx, package in enumerator:
end = time.time()
if flip_eval:
(imgs0, pids, camids, paths), (imgs1, _, _, _) = package
if use_gpu:
imgs0, imgs1 = imgs0.cuda(), imgs1.cuda()
features = (model(imgs0)[0] + model(imgs1)[0]) / 2.0
# print(features.size())
else:
(imgs, pids, camids, paths) = package
if use_gpu:
imgs = imgs.cuda()
features = model(imgs)[0]
batch_time.update(time.time() - end)
features = features.data.cpu()
qf.append(features)
q_pids.extend(pids)
q_camids.extend(camids)
q_paths.extend(paths)
qf = torch.cat(qf, 0)
q_pids = np.asarray(q_pids)
q_camids = np.asarray(q_camids)
print("Extracted features for query set, obtained {}-by-{} matrix".
format(qf.size(0), qf.size(1)))
gf, g_pids, g_camids, g_paths = [], [], [], []
if flip_eval:
enumerator = enumerate(zip(galleryloader[0], galleryloader[1]))
else:
enumerator = enumerate(galleryloader[0])
for batch_idx, package in enumerator:
end = time.time()
if flip_eval:
(imgs0, pids, camids, paths), (imgs1, _, _, _) = package
if use_gpu:
imgs0, imgs1 = imgs0.cuda(), imgs1.cuda()
features = (model(imgs0)[0] + model(imgs1)[0]) / 2.0
# print(features.size())
else:
(imgs, pids, camids, _) = package
if use_gpu:
imgs = imgs.cuda()
features = model(imgs)[0]
batch_time.update(time.time() - end)
features = features.data.cpu()
gf.append(features)
g_pids.extend(pids)
g_camids.extend(camids)
g_paths.extend(paths)
gf = torch.cat(gf, 0)
g_pids = np.asarray(g_pids)
g_camids = np.asarray(g_camids)
print("Extracted features for gallery set, obtained {}-by-{} matrix".
format(gf.size(0), gf.size(1)))
if os.environ.get('save_feat'):
import scipy.io as io
io.savemat(
os.environ.get('save_feat'), {
'q': qf.data.numpy(),
'g': gf.data.numpy(),
'qt': q_pids,
'gt': g_pids
})
# return
print("==> BatchTime(s)/BatchSize(img): {:.3f}/{}".format(
batch_time.avg, args.test_batch_size))
m, n = qf.size(0), gf.size(0)
distmat = torch.pow(qf, 2).sum(dim=1, keepdim=True).expand(m, n) + \
torch.pow(gf, 2).sum(dim=1, keepdim=True).expand(n, m).t()
distmat.addmm_(1, -2, qf, gf.t())
distmat = distmat.numpy()
if os.environ.get('distmat'):
import scipy.io as io
io.savemat(os.environ.get('distmat'), {
'distmat': distmat,
'qp': q_paths,
'gp': g_paths
})
print("Computing CMC and mAP")
cmc, mAP = evaluate_rank(distmat,
q_pids,
g_pids,
q_camids,
g_camids,
use_metric_cuhk03=args.use_metric_cuhk03)
print("Results ----------")
print("mAP: {:.2%}".format(mAP))
print("CMC curve")
for r in ranks:
print("Rank-{:<3}: {:.2%}".format(r, cmc[r - 1]))
print("------------------")
if return_distmat:
return distmat
return cmc[0]
def test(model,
queryloader,
galleryloader,
pool,
use_gpu,
ranks=(1, 5, 10, 20),
return_distmat=False):
global mAP
batch_time = AverageMeter()
model.eval()
with torch.no_grad():
qf, q_pids, q_camids = [], [], []
for batch_idx, (imgs, pids, camids, adj) in enumerate(queryloader):
if use_gpu:
imgs, adj = imgs.cuda(), adj.cuda()
if args.test_sample in ['dense', 'skipdense']:
b, n, s, c, h, w = imgs.size()
imgs = imgs.view(b * n, s, c, h, w)
adj = adj.view(b * n, adj.size(-1), adj.size(-1))
else:
n, s, c, h, w = imgs.size()
end = time.time()
features = model(imgs, adj)
batch_time.update(time.time() - end)
if args.test_sample in ['dense', 'skipdense']:
features = features.view(n, 1, -1)
if pool == 'avg':
features = torch.mean(features, 0)
else:
features, _ = torch.max(features, 0)
features = features.data.cpu()
qf.append(features)
q_pids.extend(pids.numpy())
q_camids.extend(camids.numpy())
qf = torch.cat(qf, 0)
q_pids = np.asarray(q_pids)
q_camids = np.asarray(q_camids)
print("Extracted features for query set, obtained {}-by-{} matrix".
format(qf.size(0), qf.size(1)))
gf, g_pids, g_camids = [], [], []
for batch_idx, (imgs, pids, camids, adj) in enumerate(galleryloader):
if use_gpu:
imgs, adj = imgs.cuda(), adj.cuda()
if args.test_sample in ['dense', 'skipdense']:
b, n, s, c, h, w = imgs.size()
imgs = imgs.view(b * n, s, c, h, w)
adj = adj.view(b * n, adj.size(-1), adj.size(-1))
else:
n, s, c, h, w = imgs.size()
end = time.time()
features = model(imgs, adj)
batch_time.update(time.time() - end)
if args.test_sample in ['dense', 'skipdense']:
features = features.view(n, 1, -1)
if pool == 'avg':
features = torch.mean(features, 0)
else:
features, _ = torch.max(features, 0)
features = features.data.cpu()
gf.append(features)
g_pids.extend(pids.numpy())
g_camids.extend(camids.numpy())
gf = torch.cat(gf, 0)
g_pids = np.asarray(g_pids)
g_camids = np.asarray(g_camids)
print("Extracted features for gallery set, obtained {}-by-{} matrix".
format(gf.size(0), gf.size(1)))
print("==> BatchTime(s)/BatchSize(img): {:.3f}/{}".format(
batch_time.avg, args.test_batch * args.seq_len))
print('Computing distance matrix with metric={} ...'.format(
args.dist_metric))
distmat = metrics.compute_distance_matrix(qf, gf, args.dist_metric)
distmat = distmat.numpy()
if args.re_rank:
print('Applying person re-ranking ...')
distmat_qq = metrics.compute_distance_matrix(qf, qf, args.dist_metric)
distmat_gg = metrics.compute_distance_matrix(gf, gf, args.dist_metric)
distmat = re_ranking(distmat, distmat_qq, distmat_gg)
print("Computing CMC and mAP")
cmc, mAP = metrics.evaluate_rank(distmat,
q_pids,
g_pids,
q_camids,
g_camids,
use_metric_mars=True)
print("Results ----------")
print("mAP: {:.2%}".format(mAP))
print("CMC curve")
for r in ranks:
print("Rank-{:<3}: {:.2%}".format(r, cmc[r - 1]))
print("------------------")
if return_distmat:
return distmat
return cmc[0], mAP
def _evaluate(self,
epoch,
dataset_name='',
queryloader=None,
galleryloader=None,
dist_metric='euclidean',
normalize_feature=False,
visrank=False,
visrankactiv=False,
visrank_topk=10,
save_dir='',
use_metric_cuhk03=False,
ranks=[1, 5, 10, 20],
rerank=False,
visrankactivthr=False,
maskthr=0.7,
visdrop=False,
visdroptype='random'):
batch_time = AverageMeter()
print('Extracting features from query set ...')
qf, qa, q_pids, q_camids, qm = [], [], [], [], [
] # query features, query activations, query person IDs, query camera IDs and image drop masks
for _, data in enumerate(queryloader):
imgs, pids, camids = self._parse_data_for_eval(data)
if self.use_gpu:
imgs = imgs.cuda()
end = time.time()
features = self._extract_features(imgs)
activations = self._extract_activations(imgs)
dropmask = self._extract_drop_masks(imgs, visdrop, visdroptype)
batch_time.update(time.time() - end)
features = features.data.cpu()
qf.append(features)
qa.append(torch.Tensor(activations))
qm.append(torch.Tensor(dropmask))
q_pids.extend(pids)
q_camids.extend(camids)
qf = torch.cat(qf, 0)
qm = torch.cat(qm, 0)
qa = torch.cat(qa, 0)
q_pids = np.asarray(q_pids)
q_camids = np.asarray(q_camids)
print('Done, obtained {}-by-{} matrix'.format(qf.size(0), qf.size(1)))
print('Extracting features from gallery set ...')
gf, ga, g_pids, g_camids, gm = [], [], [], [], [
] # gallery features, gallery activations, gallery person IDs, gallery camera IDs and image drop masks
end = time.time()
for _, data in enumerate(galleryloader):
imgs, pids, camids = self._parse_data_for_eval(data)
if self.use_gpu:
imgs = imgs.cuda()
end = time.time()
features = self._extract_features(imgs)
activations = self._extract_activations(imgs)
dropmask = self._extract_drop_masks(imgs, visdrop, visdroptype)
batch_time.update(time.time() - end)
features = features.data.cpu()
gf.append(features)
ga.append(torch.Tensor(activations))
gm.append(torch.Tensor(dropmask))
g_pids.extend(pids)
g_camids.extend(camids)
gf = torch.cat(gf, 0)
gm = torch.cat(gm, 0)
ga = torch.cat(ga, 0)
g_pids = np.asarray(g_pids)
g_camids = np.asarray(g_camids)
print('Done, obtained {}-by-{} matrix'.format(gf.size(0), gf.size(1)))
print('Speed: {:.4f} sec/batch'.format(batch_time.avg))
if normalize_feature:
print('Normalzing features with L2 norm ...')
qf = F.normalize(qf, p=2, dim=1)
gf = F.normalize(gf, p=2, dim=1)
print(
'Computing distance matrix with metric={} ...'.format(dist_metric))
distmat = metrics.compute_distance_matrix(qf, gf, dist_metric)
distmat = distmat.numpy()
#always show results without re-ranking first
print('Computing CMC and mAP ...')
cmc, mAP = metrics.evaluate_rank(distmat,
q_pids,
g_pids,
q_camids,
g_camids,
use_metric_cuhk03=use_metric_cuhk03)
print('** Results **')
print('mAP: {:.1%}'.format(mAP))
print('CMC curve')
for r in ranks:
print('Rank-{:<3}: {:.1%}'.format(r, cmc[r - 1]))
if rerank:
print('Applying person re-ranking ...')
distmat_qq = metrics.compute_distance_matrix(qf, qf, dist_metric)
distmat_gg = metrics.compute_distance_matrix(gf, gf, dist_metric)
distmat = re_ranking(distmat, distmat_qq, distmat_gg)
print('Computing CMC and mAP ...')
cmc, mAP = metrics.evaluate_rank(
distmat,
q_pids,
g_pids,
q_camids,
g_camids,
use_metric_cuhk03=use_metric_cuhk03)
print('** Results with Re-Ranking**')
print('mAP: {:.1%}'.format(mAP))
print('CMC curve')
for r in ranks:
print('Rank-{:<3}: {:.1%}'.format(r, cmc[r - 1]))
if visrank:
visualize_ranked_results(
distmat,
self.datamanager.return_testdataset_by_name(dataset_name),
self.datamanager.data_type,
width=self.datamanager.width,
height=self.datamanager.height,
save_dir=osp.join(save_dir, 'visrank_' + dataset_name),
topk=visrank_topk)
if visrankactiv:
visualize_ranked_activation_results(
distmat,
qa,
ga,
self.datamanager.return_testdataset_by_name(dataset_name),
self.datamanager.data_type,
width=self.datamanager.width,
height=self.datamanager.height,
save_dir=osp.join(save_dir, 'visrankactiv_' + dataset_name),
topk=visrank_topk)
if visrankactivthr:
visualize_ranked_threshold_activation_results(
distmat,
qa,
ga,
self.datamanager.return_testdataset_by_name(dataset_name),
self.datamanager.data_type,
width=self.datamanager.width,
height=self.datamanager.height,
save_dir=osp.join(save_dir, 'visrankactivthr_' + dataset_name),
topk=visrank_topk,
threshold=maskthr)
if visdrop:
visualize_ranked_mask_activation_results(
distmat,
qa,
ga,
qm,
gm,
self.datamanager.return_testdataset_by_name(dataset_name),
self.datamanager.data_type,
width=self.datamanager.width,
height=self.datamanager.height,
save_dir=osp.join(
save_dir, 'visdrop_{}_{}'.format(visdroptype,
dataset_name)),
topk=visrank_topk)
return cmc[0]
def _evaluate(self,
epoch,
dataset_name='',
queryloader=None,
galleryloader=None,
dist_metric='euclidean',
normalize_feature=False,
visrank=False,
visrank_topk=20,
save_dir='',
use_metric_cuhk03=False,
ranks=[1, 5, 10, 20],
rerank=False,
load_pose=False,
part_score=False):
batch_time = AverageMeter()
self.model.eval()
print('Extracting features from query set ...')
qf, q_pids, q_camids = [], [], [
] # query features, query person IDs and query camera IDs
q_score = []
for batch_idx, data in enumerate(queryloader):
if load_pose:
imgs, pids, camids, pose = self._parse_data_for_eval(data)
else:
imgs, pids, camids = self._parse_data_for_eval(data)
if self.use_gpu:
imgs = imgs.cuda()
end = time.time()
if load_pose:
if part_score:
features, score = self._extract_features(imgs, pose)
score = score.data.cpu()
q_score.append(score)
else:
features = self._extract_features(imgs, pose)
else:
features = self._extract_features(imgs)
batch_time.update(time.time() - end)
features = features.data.cpu()
qf.append(features)
q_pids.extend(pids)
q_camids.extend(camids)
qf = torch.cat(qf, 0)
q_pids = np.asarray(q_pids)
q_camids = np.asarray(q_camids)
if part_score:
q_score = torch.cat(q_score)
print('Done, obtained {}-by-{} matrix'.format(qf.size(0), qf.size(1)))
print('Extracting features from gallery set ...')
gf, g_pids, g_camids = [], [], [
] # gallery features, gallery person IDs and gallery camera IDs
g_score = []
end = time.time()
for batch_idx, data in enumerate(galleryloader):
if load_pose:
imgs, pids, camids, pose = self._parse_data_for_eval(data)
else:
imgs, pids, camids = self._parse_data_for_eval(data)
if self.use_gpu:
imgs = imgs.cuda()
end = time.time()
if load_pose:
# if part_score:
if part_score:
features, score = self._extract_features(imgs, pose)
score = score.data.cpu()
g_score.append(score)
else:
features = self._extract_features(imgs, pose)
else:
features = self._extract_features(imgs)
batch_time.update(time.time() - end)
features = features.data.cpu()
gf.append(features)
g_pids.extend(pids)
g_camids.extend(camids)
gf = torch.cat(gf, 0)
g_pids = np.asarray(g_pids)
g_camids = np.asarray(g_camids)
if part_score:
g_score = torch.cat(g_score)
print('Done, obtained {}-by-{} matrix'.format(gf.size(0), gf.size(1)))
print('Speed: {:.4f} sec/batch'.format(batch_time.avg))
if normalize_feature:
print('Normalzing features with L2 norm ...')
qf = F.normalize(qf, p=2, dim=1)
gf = F.normalize(gf, p=2, dim=1)
print(
'Computing distance matrix with metric={} ...'.format(dist_metric))
if part_score:
distmat = metrics.compute_weight_distance_matrix(
qf, gf, q_score, g_score, dist_metric)
else:
distmat = metrics.compute_distance_matrix(qf, gf, dist_metric)
distmat = distmat.numpy()
if rerank:
print('Applying person re-ranking ...')
distmat_qq = metrics.compute_distance_matrix(qf, qf, dist_metric)
distmat_gg = metrics.compute_distance_matrix(gf, gf, dist_metric)
distmat = re_ranking(distmat, distmat_qq, distmat_gg)
print('Computing CMC and mAP ...')
cmc, mAP = metrics.evaluate_rank(distmat,
q_pids,
g_pids,
q_camids,
g_camids,
use_metric_cuhk03=use_metric_cuhk03)
print('** Results **')
print('mAP: {:.1%}'.format(mAP))
print('CMC curve')
for r in ranks:
print('Rank-{:<3}: {:.1%}'.format(r, cmc[r - 1]))
if visrank:
visualize_ranked_results(
distmat,
self.datamanager.return_testdataset_by_name(dataset_name),
save_dir=osp.join(save_dir, 'visrank-' + str(epoch + 1),
dataset_name),
topk=visrank_topk)
return cmc[0]
