pytorch 1.0 version - error on custom COCO dataset

running a custom dataset, seeing the following

Appending horizontally-flipped training examples...
coco_try_20190419_train gt roidb loaded from /media/latest_mount/Dev/git/EMCP/faster-rcnn.pytorch/data/cache/coco_try_20190419_train_gt_roidb.pkl
done
Preparing training data...
Backend TkAgg is interactive backend. Turning interactive mode on.
done
loading annotations into memory.../media/latest_mount/Dev/git/EMCP/faster-rcnn.pytorch/data/coco_try_20190419/annotations/instances_train_20190419.json at 1556370591.791137
Done (t=0.00s)
creating index...
index created!
before filtering, there are 254 images...
after filtering, there are 254 images...
254 roidb entries
Loading pretrained weights from data/pretrained_model/resnet101_caffe.pth
Traceback (most recent call last):
  File "/home/emcp/Tools/pycharm-community-2018.2.4/helpers/pydev/pydevd.py", line 1741, in <module>
    main()
  File "/home/emcp/Tools/pycharm-community-2018.2.4/helpers/pydev/pydevd.py", line 1735, in main
    globals = debugger.run(setup['file'], None, None, is_module)
  File "/home/emcp/Tools/pycharm-community-2018.2.4/helpers/pydev/pydevd.py", line 1135, in run
    pydev_imports.execfile(file, globals, locals)  # execute the script
  File "/home/emcp/Tools/pycharm-community-2018.2.4/helpers/pydev/_pydev_imps/_pydev_execfile.py", line 18, in execfile
    exec(compile(contents+"\n", file, 'exec'), glob, loc)
  File "/media/latest_mount/Dev/git/EMCP/faster-rcnn.pytorch/trainval_net.py", line 322, in <module>
    data = next(data_iter)
  File "/home/emcp/anaconda3/envs/pytorch1_py36/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 615, in __next__
    batch = self.collate_fn([self.dataset[i] for i in indices])
  File "/home/emcp/anaconda3/envs/pytorch1_py36/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 615, in <listcomp>
    batch = self.collate_fn([self.dataset[i] for i in indices])
  File "/media/latest_mount/Dev/git/EMCP/faster-rcnn.pytorch/lib/roi_data_layer/roibatchLoader.py", line 67, in __getitem__
    blobs = get_minibatch(minibatch_db, self._num_classes)
  File "/media/latest_mount/Dev/git/EMCP/faster-rcnn.pytorch/lib/roi_data_layer/minibatch.py", line 30, in get_minibatch
    im_blob, im_scales = _get_image_blob(roidb, random_scale_inds)
  File "/media/latest_mount/Dev/git/EMCP/faster-rcnn.pytorch/lib/roi_data_layer/minibatch.py", line 79, in _get_image_blob
    cfg.TRAIN.MAX_SIZE)
  File "/media/latest_mount/Dev/git/EMCP/faster-rcnn.pytorch/lib/model/utils/blob.py", line 39, in prep_im_for_blob
    im -= pixel_means
ValueError: operands could not be broadcast together with shapes (1880,2880,4) (1,1,3) (1880,2880,4) 

[EMCP]

you have to change it in two places… once in the training python file, and one in the demo…

if you detect 4 channels, and you’re sure that the alpha channel is unused, go ahead and transform the image using python… to only 3 channels

https://github.com/jwyang/faster-rcnn.pytorch/blob/master/demo.py#L115-L118

  im_orig = im.astype(np.float32, copy=True)
  if im_orig.shape[2] == 4:
      # in the event you have an image with alpha channels, drop it for now
      im_orig = im_orig[:, :, :3]
  im_orig -= cfg.PIXEL_MEANS	  im_orig -= cfg.PIXEL_MEANS

[miaozhang0525]

Using custom coco dataset, errors happend

after filtering, there are 0 images...
0 roidb entries
Loading pretrained weights from data/pretrained_model/resnet101_caffe.pth
Traceback (most recent call last):
  File "trainval_net.py", line 373, in <module>
    save_name = os.path.join(output_dir, 'faster_rcnn_{}_{}_{}.pth'.format(args.session, epoch, step))
NameError: name 'step' is not defined

could you drop your custom data here code because you have to modify

Have you guys solved it??