0 instance detected

[frischzenger]

i used COCO-InstanceSegmentation/mask_cascade_rcnn_ResNeSt_200_FPN_syncBN_all_tricks_3x.yaml this model but 0 instance detected, this is test [image:](https://timgsa.baidu.com/timg?image&quality=80&size=b9999_10000&sec=1607682904616&di=6b5afd1e92283f9170ebf4287c308263&imgtype=0&src=http%3A%2F%2Fimg11.360buyimg.com%2Fn1%2Fs350x449_jfs%2Ft2686%2F169%2F35030773%2F235194%2F3428fb86%2F56fcc929N2fb70bda.jpg%2521cc_350x449.jpg)

[frischzenger]

and I test many other pictures, but also 0 instance detected

[frischzenger]

and it hints some mode parameters are not in the checkpoint

[frischzenger]

i also test other COCO-InstanceSegmentation models, but nothing detected, what's the problem?

[frischzenger]

and i test official detectron2 it works well

[NVukobrat]

Hi @frischzenger, is there some specific reason you chose /mask_cascade_rcnn_ResNeSt_200_FPN_syncBN_all_tricks_3x.yaml?

The problem can possibly be with the wrong pair of the configuration files and/or weights. Can you post the full output?

Also, can you try to run following configuration: mask_cascade_rcnn_ResNeSt_200_FPN_dcn_syncBN_all_tricks_3x.yaml

with these weights: https://s3.us-west-1.wasabisys.com/resnest/detectron/mask_cascade_rcnn_ResNeSt_200_FPN_dcn_syncBN_all_tricks_3x-e1901134.pth

[mcwoojcik]

Hello - I obtained a similar problem (it is similar for all ResNeSt variations 50, 101, 200 using cpu because CUDA out of memory):

python demo.py --config-file ../configs/COCO-InstanceSegmentation/mask_cascade_rcnn_ResNeSt_101_FPN_syncBN_1x.yaml --input /home/kaczor/crop_test_yolact/multipla_/2b46bab5-52de-47b0-07f1-08d89a8f02c4.jpg --opts MODEL.DEVICE cpu

[01/19 16:57:05 detectron2]: Arguments: Namespace(confidence_threshold=0.5, config_file='../configs/COCO-InstanceSegmentation/mask_cascade_rcnn_ResNeSt_101_FPN_syncBN_1x.yaml', input=['/home/kaczor/crop_test_yolact/multipla_/2b46bab5-52de-47b0-07f1-08d89a8f02c4.jpg'], opts=['MODEL.DEVICE', 'cpu'], output=None, video_input=None, webcam=False)
[01/19 16:57:07 fvcore.common.checkpoint]: Loading checkpoint from https://s3.us-west-1.wasabisys.com/resnest/detectron/resnest101_detectron-486f69a8.pth
[01/19 16:57:07 fvcore.common.file_io]: Downloading https://s3.us-west-1.wasabisys.com/resnest/detectron/resnest101_detectron-486f69a8.pth ...
[01/19 16:57:07 fvcore.common.download]: Downloading from https://s3.us-west-1.wasabisys.com/resnest/detectron/resnest101_detectron-486f69a8.pth ...
resnest101_detectron-486f69a8.pth: 186MB [02:20, 1.32MB/s]                      
[01/19 16:59:28 fvcore.common.download]: Successfully downloaded /home/kaczor/.torch/fvcore_cache/resnest/detectron/resnest101_detectron-486f69a8.pth. 185563744 bytes.
[01/19 16:59:28 fvcore.common.file_io]: URL https://s3.us-west-1.wasabisys.com/resnest/detectron/resnest101_detectron-486f69a8.pth cached in /home/kaczor/.torch/fvcore_cache/resnest/detectron/resnest101_detectron-486f69a8.pth
[01/19 16:59:28 fvcore.common.checkpoint]: Some model parameters are not in the checkpoint:
  roi_heads.box_head.2.conv2.weight
  backbone.fpn_lateral2.norm.{running_var, bias, running_mean, weight}
  roi_heads.box_head.2.conv4.norm.{bias, weight, running_mean, running_var}
  backbone.fpn_output3.norm.{bias, running_mean, weight, running_var}
  roi_heads.box_predictor.2.bbox_pred.{weight, bias}
  proposal_generator.rpn_head.conv.{weight, bias}
  roi_heads.box_head.0.conv4.norm.{running_mean, running_var, weight, bias}
  backbone.fpn_output5.norm.{running_var, running_mean, bias, weight}
  roi_heads.box_head.0.conv2.norm.{running_var, running_mean, bias, weight}
  proposal_generator.rpn_head.anchor_deltas.{bias, weight}
  roi_heads.mask_head.predictor.{bias, weight}
  roi_heads.mask_head.mask_fcn4.norm.{bias, running_var, weight, running_mean}
  roi_heads.box_head.2.conv3.norm.{running_mean, weight, bias, running_var}
  roi_heads.box_head.0.conv3.weight
  roi_heads.box_head.1.conv4.norm.{running_mean, bias, weight, running_var}
  backbone.fpn_lateral5.weight
  backbone.fpn_output4.norm.{running_var, weight, bias, running_mean}
  roi_heads.box_predictor.0.bbox_pred.{bias, weight}
  backbone.fpn_lateral3.norm.{running_mean, bias, weight, running_var}
  roi_heads.box_head.1.fc1.{weight, bias}
  roi_heads.box_predictor.0.cls_score.{bias, weight}
  backbone.fpn_lateral5.norm.{weight, running_mean, running_var, bias}
  roi_heads.box_head.1.conv2.norm.{running_var, bias, weight, running_mean}
  roi_heads.box_head.2.conv2.norm.{running_mean, bias, weight, running_var}
  roi_heads.mask_head.mask_fcn1.weight
  roi_heads.box_head.1.conv3.norm.{running_mean, running_var, bias, weight}
  roi_heads.mask_head.mask_fcn3.norm.{running_var, weight, bias, running_mean}
  proposal_generator.anchor_generator.cell_anchors.{1, 0, 2, 3, 4}
  roi_heads.mask_head.mask_fcn2.weight
  roi_heads.box_head.0.conv3.norm.{weight, running_mean, bias, running_var}
  roi_heads.box_predictor.1.cls_score.{weight, bias}
  roi_heads.mask_head.deconv.{bias, weight}
  roi_heads.box_head.1.conv1.norm.{running_mean, running_var, weight, bias}
  roi_heads.box_head.0.conv1.norm.{weight, bias, running_mean, running_var}
  roi_heads.mask_head.mask_fcn1.norm.{weight, bias, running_var, running_mean}
  roi_heads.mask_head.mask_fcn2.norm.{weight, running_var, running_mean, bias}
  proposal_generator.rpn_head.objectness_logits.{weight, bias}
  backbone.fpn_lateral3.weight
  roi_heads.box_head.1.conv3.weight
  roi_heads.box_head.2.conv1.weight
  backbone.fpn_output2.norm.{running_mean, running_var, weight, bias}
  roi_heads.box_head.2.conv1.norm.{bias, running_mean, running_var, weight}
  backbone.fpn_lateral4.norm.{weight, running_mean, bias, running_var}
  roi_heads.box_predictor.2.cls_score.{bias, weight}
  roi_heads.box_head.1.conv2.weight
  roi_heads.box_head.2.conv4.weight
  roi_heads.box_head.1.conv4.weight
  roi_heads.box_head.0.fc1.{weight, bias}
  backbone.fpn_lateral4.weight
  roi_heads.box_head.2.conv3.weight
  roi_heads.box_head.0.conv4.weight
  backbone.fpn_output3.weight
  backbone.fpn_output5.weight
  backbone.fpn_output2.weight
  backbone.fpn_lateral2.weight
  roi_heads.mask_head.mask_fcn3.weight
  roi_heads.box_head.0.conv2.weight
  roi_heads.box_head.0.conv1.weight
  roi_heads.box_head.1.conv1.weight
  backbone.fpn_output4.weight
  roi_heads.box_predictor.1.bbox_pred.{weight, bias}
  roi_heads.mask_head.mask_fcn4.weight
  roi_heads.box_head.2.fc1.{weight, bias}
/home/kaczor/detectron2-ResNeSt/detectron2/modeling/roi_heads/fast_rcnn.py:107: UserWarning: This overload of nonzero is deprecated:
	nonzero()
Consider using one of the following signatures instead:
	nonzero(*, bool as_tuple) (Triggered internally at  /pytorch/torch/csrc/utils/python_arg_parser.cpp:882.)
  filter_inds = filter_mask.nonzero()
[01/19 16:59:45 detectron2]: /home/kaczor/crop_test_yolact/multipla_/2b46bab5-52de-47b0-07f1-08d89a8f02c4.jpg: detected 0 instances in 16.87s

I installed detectron2 using detectron2-ResNeSt repo by python -m pip install -e . command in detectron2-ResNeSt directory

OS: Ubuntu 18.04.5 LTS x64 PyTorch: 1.7.0+cu110 CUDA: 11.1 Python: 3.6.9 Run on cpu because CUDA out of memory RuntimeError by --opts MODEL.DEVICE cpu option

I'll be grateful for any hints about the possible reason or for the solution which will help me run inference successfully. Best regards :)

[mcwoojcik]

I have the same problem using google COLAB - maybe I used wrong command in demo to run?