I wanted to evaluate a video-text alignment model on the ssv2-template and ssv2-label dataset. I downloaded the videos from qualcomm website and extracted all the videos from their zip files.
Following this, i was running my code on the extracted webm videos and found that 1989 videos out of 2088 videos are corrupted. The code to load video is here:
from decord import VideoReader
import numpy as np
from PIL import Image
def get_index(num_frames, num_segments):
seg_size = float(num_frames - 1) / num_segments
start = int(seg_size / 2)
offsets = np.array([
start + int(np.round(seg_size * idx)) for idx in range(num_segments)
])
return offsets
def load_video(path, num_frames=4):
vr = VideoReader(path, height=224, width=224)
total_frames = len(vr)
frame_indices = get_index(total_frames, num_frames)
images_group = list()
for frame_index in frame_indices:
img = Image.fromarray(vr[frame_index].asnumpy()).convert('RGB')
images_group.append(img)
return images_group
Have you seen this corruption issue before? If possible, can you share your 2088 val small videos for SSv2 with me on my email?
Hi @jayleicn ,
I wanted to evaluate a video-text alignment model on the ssv2-template and ssv2-label dataset. I downloaded the videos from qualcomm website and extracted all the videos from their zip files.
Following this, i was running my code on the extracted webm videos and found that 1989 videos out of 2088 videos are corrupted. The code to load video is here:
Have you seen this corruption issue before? If possible, can you share your 2088 val small videos for SSv2 with me on my email?