Skip to content

Open-Sora: Democratizing Efficient Video Production for All

Python

fvd debug

Shen-Chenhui76bbe322640c
diff --git a/eval/cal_fvd.py b/eval/cal_fvd.py
index 1f1a980..949b773 100644
--- a/eval/cal_fvd.py
+++ b/eval/cal_fvd.py
@@ -2,7 +2,7 @@ import numpy as np
import torch
from tqdm import tqdm
-def trans(x):
+def trans(x): # requires video to be BCTHW
# if greyscale images add channel
if x.shape[-3] == 1:
x = x.repeat(1, 1, 3, 1, 1)
@@ -12,6 +12,7 @@ def trans(x):
return x
+
def calculate_fvd(videos1, videos2, device, method='styleganv'):
if method == 'styleganv':
@@ -39,6 +40,7 @@ def calculate_fvd(videos1, videos2, device, method='styleganv'):
fvd_results = {}
+
# for calculate FVD, each clip_timestamp must >= 10
for clip_timestamp in tqdm(range(10, videos1.shape[-3]+1)):
@@ -47,10 +49,12 @@ def calculate_fvd(videos1, videos2, device, method='styleganv'):
videos_clip1 = videos1[:, :, : clip_timestamp]
videos_clip2 = videos2[:, :, : clip_timestamp]
+ breakpoint()
# get FVD features
feats1 = get_fvd_feats(videos_clip1, i3d=i3d, device=device)
feats2 = get_fvd_feats(videos_clip2, i3d=i3d, device=device)
-
+
+ breakpoint()
# calculate FVD when timestamps[:clip]
fvd_results[clip_timestamp] = frechet_distance(feats1, feats2)
diff --git a/eval/eval_common_metric.py b/eval/eval_common_metric.py
index abca5fa..4e03b4a 100644
--- a/eval/eval_common_metric.py
+++ b/eval/eval_common_metric.py
@@ -42,11 +42,11 @@ from torchvision.transforms import Lambda, Compose
from torchvision.transforms._transforms_video import CenterCropVideo
import sys
sys.path.append(".")
-from .cal_lpips import calculate_lpips
-from .cal_fvd import calculate_fvd
-from .cal_psnr import calculate_psnr
-from .cal_flolpips import calculate_flolpips
-from .cal_ssim import calculate_ssim
+# from cal_lpips import calculate_lpips
+from cal_fvd import calculate_fvd
+from cal_psnr import calculate_psnr
+# from cal_flolpips import calculate_flolpips
+from cal_ssim import calculate_ssim
try:
from tqdm import tqdm
@@ -146,6 +146,7 @@ def calculate_common_metric(args, dataloader, device):
assert real_videos.shape[2] == generated_videos.shape[2]
if args.metric == 'fvd':
tmp_list = list(calculate_fvd(real_videos, generated_videos, args.device, method=args.fvd_method)['value'].values())
+ print("fvd list:", tmp_list)
elif args.metric == 'ssim':
tmp_list = list(calculate_ssim(real_videos, generated_videos)['value'].values())
elif args.metric == 'psnr':
diff --git a/eval/fvd/styleganv/fvd.py b/eval/fvd/styleganv/fvd.py
index 3043a2a..e1e9fde 100644
--- a/eval/fvd/styleganv/fvd.py
+++ b/eval/fvd/styleganv/fvd.py
@@ -10,6 +10,7 @@ def load_i3d_pretrained(device=torch.device('cpu')):
i3D_WEIGHTS_URL = "https://www.dropbox.com/s/ge9e5ujwgetktms/i3d_torchscript.pt"
filepath = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'i3d_torchscript.pt')
print(filepath)
+ breakpoint()
if not os.path.exists(filepath):
print(f"preparing for download {i3D_WEIGHTS_URL}, you can download it by yourself.")
os.system(f"wget {i3D_WEIGHTS_URL} -O {filepath}")