"""Main script for Youtube-8M feature extractor."""
import misc.config as cfg
from misc.utils import concat_feat, get_dataloader, make_cuda, make_variable
from models import PCAWrapper, inception_v3
if __name__ == '__main__':
# init models and data loader
model = make_cuda(
inception_v3(pretrained=True, transform_input=True, extract_feat=True))
pca = PCAWrapper(n_components=cfg.n_components)
model.eval()
# data loader for frames in ingle video
# data_loader = get_dataloader(dataset="VideoFrame",
# path=cfg.video_file,
# num_frames=cfg.num_frames,
# batch_size=cfg.batch_size)
# data loader for frames decoded from several videos
data_loader = get_dataloader(dataset="FrameImage",
path=cfg.frame_root,
batch_size=cfg.batch_size)
# extract features by inception_v3
feats = None
for step, frames in enumerate(data_loader):
print("extracting feature [{}/{}]".format(step + 1, len(data_loader)))
feat = model(make_variable(frames))
feats = concat_feat(feats, feat.data.cpu())
# recude dimensions by PCA
QUANTIZE_MAX_VAL)
# - convert to 8-bit in range [0.0, 255.0]
quantized_embeddings = (
(clipped_embeddings - QUANTIZE_MIN_VAL) *
(255.0 /
(QUANTIZE_MAX_VAL - QUANTIZE_MIN_VAL)))
# - cast 8-bit float to uint8
quantized_embeddings = quantized_embeddings.astype(np.uint8)
return quantized_embeddings
if __name__ == '__main__':
# init Inception v3 model
model = make_cuda(inception_v3(pretrained=True,
model_path=cfg.inception_v3_model,
transform_input=True,
extract_feat=True))
model.eval()
# init PCA model
pca = PCAWrapper(n_components=cfg.n_components,
batch_size=cfg.pca_batch_size)
pca.load_params(filepath=cfg.pca_model)
subfolders = list_folders(cfg.dataset_path)
for subfolder in subfolders:
print("current folder: {}".format(subfolder))
# data loader for frames in single video
data_loader = get_dataloader(dataset="FrameImage",
path=os.path.join(subfolder, 'frames'),
frame_num =cfg.frame_num,
