dawn-lx / vidvrd-tracklets Goto Github PK

Hello,
I failed to follow the mega.pytorch.INSTALL.md due to the CUDA version being incompatible with the Pytorch version. I exactly followed the INSTALL.md step by step, and I did install pytorch=1.3.0 and cudatoolkit=10.0 successfully. But, I still failed to install the apex.

So, I jumped back to your repo. May I please ask you if you are still trying to release the 'modification_details.md' later on or all we need to do is just to replace the official 'mega.pytorch/mega_core' with your modified 'MEGA/mega_core' ?

Thank yous so much for your time! Looking forward to your response soon!
Cheers!

Near the line 251 of deepSORT/deep_sort_app_v2.py:

for i, track in enumerate(tracker.tracks):
    if not track.is_confirmed() or track.time_since_update > 1:
        continue
    bbox = track.to_tlwh()
    track_res = [frame_idx, track.track_id, bbox[0], bbox[1], bbox[2],
            bbox[3]]
    if i not in track2det.keys():
        results.append(track_res)
        continue
    track_res += [detections[track2det[i]].confidence]
    track_res += [detections[track2det[i]].label]
    track_res += list(detections[track2det[i]].tlwh)
    track_res += list(detections[track2det[i]].feature)
    results.append(track_res)

The are 2 statements calling results.append(track_res). The first one always appends a list with length 6, but the last one is 1036, which will make np.array(results) fail:

Traceback (most recent call last):
  File "/home/chuanwise/develop/projects/VidVRD-tracklets/deepSORT/deepSORT_tracking_v2.py", line 94, in <module>
    run(
  File "/home/chuanwise/develop/projects/VidVRD-tracklets/deepSORT/deep_sort_app_v2.py", line 244, in run
    visualizer.run(frame_callback)
  File "/home/chuanwise/develop/projects/VidVRD-tracklets/deepSORT/application_util/visualization.py", line 80, in run
    frame_callback(self, self.frame_idx)
  File "/home/chuanwise/develop/projects/VidVRD-tracklets/deepSORT/deep_sort_app_v2.py", line 185, in frame_callback
Traceback (most recent call last):
  File "/home/chuanwise/develop/projects/VidVRD-tracklets/deepSORT/deepSORT_tracking_v2.py", line 94, in <module>
    run(
  File "/home/chuanwise/develop/projects/VidVRD-tracklets/deepSORT/deep_sort_app_v2.py", line 254, in run
    res = np.array(results)
          ^^^^^^^^^^^^^^^^^
ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (389,) + inhomogeneous part.        

你好，VidOR原始数据集无法从其主页上下载，您是否有备份可以分享？
感谢！

I followed the steps of MEGA installations and everything went well, but during the detection of frame-level object proposals, I am getting this error:

"raise RuntimeError("{} is a zip archive (did you mean to use torch.jit.load()?)".format(f.name))
RuntimeError: MEGA/training_dir/COCO34ORfreq32_4gpu/model_0180000.pth is a zip archive (did you mean to use torch.jit.load()?)"

Any solution?
PS: command used:
CUDA_VISIBLE_DEVICES=0 python MEGA/tools/test_net.py --config-file MEGA/configs/MEGA/inference/VidORval_freq1_0024.yaml MODEL.WEIGHT MEGA/training_dir/COCO34ORfreq32_4gpu/model_0180000.pth OUTPUT_DIR MEGA/training_dir/COCO34ORfreq32_4gpu/inference

您好，VidOR的官网上的原始数据集下载链接失效了。请问您可以分享一下原始数据吗？比如百度网盘。万分感谢！