nicolabs/Open-Sora
1
0
Fork 0
mirror of https://github.com/hpcaitech/Open-Sora.git synced 2026-05-21 11:59:01 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

format

Browse source
This commit is contained in:
Shen-Chenhui 2024-06-10 02:12:33 +00:00
parent e852466200
commit 16866a4b32
3 changed files with 51 additions and 12 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

30
eval/vbench/tabulate_vbench_scores.py
View file

@ -65,6 +65,28 @@ DIM_WEIGHT = {
"overall consistency": 1,
}
ordered_scaled_res = [
"total score",
"quality score",
"semantic score",
"subject consistency",
"background consistency",
"temporal flickering",
"motion smoothness",
"dynamic degree",
"aesthetic quality",
"imaging quality",
"object class",
"multiple objects",
"human action",
"color",
"spatial relationship",
"scene",
"appearance style",
"temporal style",
"overall consistency",
]
def parse_args():
parser = argparse.ArgumentParser()
@ -116,9 +138,11 @@ if __name__ == "__main__":
QUALITY_WEIGHT + SEMANTIC_WEIGHT
)
formated_scaled_results = {}
for key, val in scaled_results.items():
formated_scaled_results[key] = format(val * 100, ".2f") + "%"
formated_scaled_results = {"items": []}
for key in ordered_scaled_res:
# formated_scaled_results[key] = format(val * 100, ".2f") + "%"
formated_score = format(scaled_results[key] * 100, ".2f") + "%"
formated_scaled_results["items"].append({key: formated_score})
output_file_path = os.path.join(args.score_dir, "all_results.json")
with open(output_file_path, "w") as outfile:

10
eval/vbench_i2v/calc_vbench_i2v.py
View file

@ -53,19 +53,17 @@ if __name__ == "__main__":
if args.calc_i2v:
my_VBench_I2V = VBenchI2V(torch.device("cuda"), full_info_path, output_dir)
if args.end == -1: # adjust end accordingly
args.end = len(i2v_dimensions)
for i2v_dim in i2v_dimensions[args.start : args.end]:
end = args.end if args.end != -1 else len(i2v_dimensions)
for i2v_dim in i2v_dimensions[args.start : end]:
my_VBench_I2V.evaluate(videos_path=video_path, name=i2v_dim, dimension_list=[i2v_dim], resolution="1-1")
kwargs = {}
kwargs["imaging_quality_preprocessing_mode"] = "longer" # use VBench/evaluate.py default
if args.calc_quality:
if args.end == -1: # adjust end accordingly
args.end = len(video_quality_dimensions)
my_VBench = VBench(torch.device("cuda"), full_info_path, output_dir)
for quality_dim in video_quality_dimensions[args.start : args.end]:
end = args.end if args.end != -1 else len(video_quality_dimensions)
for quality_dim in video_quality_dimensions[args.start : end]:
my_VBench.evaluate(
videos_path=video_path, name=quality_dim, dimension_list=[quality_dim], mode="vbench_standard", **kwargs
)

23
eval/vbench_i2v/tabulate_vbench_i2v_scores.py
View file

@ -46,6 +46,22 @@ NORMALIZE_DIC_I2V = {
"temporal_flickering": {"Min": 0.6293, "Max": 1.0},
}
ordered_scaled_res = [
"total score",
"i2v score",
"quality score",
"camera_motion",
"i2v_subject",
"i2v_background",
"subject_consistency",
"background_consistency",
"motion_smoothness",
"dynamic_degree",
"aesthetic_quality",
"imaging_quality",
"temporal_flickering",
]
def parse_args():
parser = argparse.ArgumentParser()
@ -100,9 +116,10 @@ if __name__ == "__main__":
I2V_QUALITY_WEIGHT + I2V_WEIGHT
)
formated_scaled_results = {}
for key, val in scaled_results.items():
formated_scaled_results[key] = format(val * 100, ".2f") + "%"
formated_scaled_results = {"item": []}
for key in ordered_scaled_res:
formated_res = format(scaled_results[key] * 100, ".2f") + "%"
formated_scaled_results["item"].append({key: formated_res})
output_file_path = os.path.join(args.score_dir, "all_results.json")
with open(output_file_path, "w") as outfile: