search

Zhongdao / Towards-Realtime-MOT

Joint Detection and Embedding for fast multi-object tracking
MIT License
2.37k stars 539 forks source link

Got error when using yolov3-tiny.conv.15 and yolov3-tiny.cfg #224

Closed mheriyanto closed 3 years ago

mheriyanto commented 3 years ago

I got error when using yolov3-tiny.conv.15 (https://pjreddie.com/media/files/yolov3-tiny.conv.15) and yolov3-tiny.cfg (https://github.com/pjreddie/darknet/blob/master/cfg/yolov3-tiny.cfg) likes below:

Traceback (most recent call last):
  File "train.py", line 298, in <module>
    opt=opt,
  File "train.py", line 111, in train
    load_darknet_weights(model, osp.join(weights_from, 'yolov3-tiny.conv.15'))
  File "/content/gdrive/My Drive/bla-bla/models.py", line 334, in load_darknet_weights
    self.seen = header[3]  # number of images seen during training
IndexError: index 3 is out of bounds for axis 0 with size 0

Actually, I modified yolov3-tiny.cfg with adding embedding_dim=512 in iine 10 and change anchors to anchors = 14,10, 27,23, 58,37, 82,81, 169,135, 319,344 in line 135 and 177.

Are there any suggestions? Thanks.

kikirizki commented 3 years ago

would you mind post your cfg file

mheriyanto commented 3 years ago

would you mind post your cfg file

Okay @kikirizki, this is yolov3-tiny.cfg.

mheriyanto commented 3 years ago

Is there any update here?

mheriyanto commented 3 years ago

I closed this issue because this is duplicate issue with https://github.com/Zhongdao/Towards-Realtime-MOT/issues/32