def test_data_path():
s = data_path()
assert isinstance(s, Path) and s != "" and os.path.isdir(
str(s)) and s.name == "data"
for p in all_val_videos_path[:NUM_OF_SAMPLES_VAL]:
class_id = LABELS.index(p.split('/')[-2])
entry = '{},{}'.format(p.replace('.mp4', ''), class_id)
filehandle.write('%s\n' % entry)
##
NUM_FRAMES = 32 # 8 or 32.
IM_SCALE = 128 # resize then crop
INPUT_SIZE = 112 # input clip size: 3 x NUM_FRAMES x 112 x 112
# video sample to download
sample_video_url = Urls.webcam_vid
# file path to save video sample
video_fpath = data_path() / "sample_video.mp4"
# prediction score threshold
SCORE_THRESHOLD = 0.01
# Averaging 5 latest clips to make video-level prediction (or smoothing)
AVERAGING_SIZE = 5
##
if BASE_MODEL == "kinetics":
BATCH_SIZE_ARRAY = [8, 1, 4, 5, 6]
MODEL_INPUT_SIZE = 32
elif BASE_MODEL == "r2plus1d_18":
BATCH_SIZE_ARRAY = [8, 1, 16, 20, 20]
MODEL_INPUT_SIZE = 16
def test__DatasetSpec_hmdb():
""" Tests DatasetSpec initialize with hmdb51 classes """
hmdb51 = _DatasetSpec(Urls.hmdb51_label_map, 51)
hmdb51.class_names
assert os.path.exists(str(data_path() / "label_map.txt"))
# models.vgg19_bn models.wrn models.wrn_22
# models.xception models.xresnet models.xresnet101
# models.xresnet152 models.xresnet18 models.xresnet34
# models.xresnet50
model = model_to_learner(models.resnet18(pretrained=True), IMAGENET_IM_SIZE)
#learn = model_to_learner(models.resnet152(pretrained=True), IMAGENET_IM_SIZE)
#learn = model_to_learner(models.xresnet152(pretrained=True), IMAGENET_IM_SIZE)
# TODO: Want to load from local copy rather than from ~/.torch? Maybe
#learn = load_learner(file="resnet18-5c106cde.pth")
#model = untar_data("resnet18-5c106cde.pth")
#learn = load_learner(model)
# TODO Handle folder of images.
for path in args.path:
if is_url(path):
imfile = os.path.join(data_path(), "temp.jpg")
urllib.request.urlretrieve(path, imfile)
else:
imfile = os.path.join(get_cmd_cwd(), path)
im = open_image(imfile, convert_mode='RGB')
# Predict the class label.
_, ind, prob = model.predict(im)
sys.stdout.write(f"{prob[ind]:.2f},{labels[ind]},{path}\n")
def test__DatasetSpec_kinetics():
""" Tests DatasetSpec initialize with kinetics classes """
kinetics = _DatasetSpec(Urls.kinetics_label_map, 400)
kinetics.class_names
assert os.path.exists(str(data_path() / "label_map.txt"))