MCPcopy Create free account
hub / github.com/UVA-Computer-Vision-Lab/OmniShotCut / parse_args

Function parse_args

inference.py:74–115  ·  view source on GitHub ↗
()

Source from the content-addressed store, hash-verified

72
73
74def parse_args():
75 parser = argparse.ArgumentParser()
76
77 parser.add_argument(
78 "--checkpoint_path",
79 type = str,
80 default = None,
81 help = "Path to local checkpoint file. Auto-downloads from HuggingFace if not set."
82 )
83 parser.add_argument(
84 "--input_video_path",
85 type = str,
86 default = "__assets__/demo_video1.mp4",
87 help = "Path to the input video path."
88 )
89 parser.add_argument(
90 "--result_store_path",
91 type = str,
92 default = "results.json",
93 help="Path to save result json."
94 )
95 parser.add_argument(
96 "--overlap_window_length",
97 type = int,
98 default = 20,
99 help = "Number of overlapped frames between adjacent inference windows."
100 )
101 parser.add_argument(
102 "--visual_store_folder_path",
103 type = str,
104 default = "demo_video_results",
105 help = "Path to save the visualization results. Set to None to disable."
106 )
107 parser.add_argument(
108 "--mode",
109 type = str,
110 default = "default",
111 choices = ["default", "clean_shot"],
112 help = "Output Mode. 'default' means all Intra and Inter label. 'clean_shot' means only General Shot Cut without transitions. "
113 )
114
115 return parser.parse_args()
116
117
118

Callers 1

inference.pyFile · 0.85

Calls

no outgoing calls

Tested by

no test coverage detected