Export yolo instance segmentation

[nik123]

I'm getting myself acquainted with instance mmyolo framework and more specifically with instance segmentation task. I was able to train the model according to the tutorial. However, it lacks the export section.

I tried to adopt export example from object detection tutorial as the following:

WEIGHTS="work_dirs/yolov5_ins_s-v61_syncbn_fast_8xb16-300e_balloon_instance/best_coco_bbox_mAP_epoch_240.pth"
                                                                                                             
python projects/easydeploy/tools/export_onnx.py \                                                            
        configs/yolov5/ins_seg/yolov5_ins_s-v61_syncbn_fast_8xb16-300e_balloon_instance.py \                 
        "${WEIGHTS}" \                                                                                       
        --work-dir work_dirs/yolov5_ins_s-v61_syncbn_fast_8xb16-300e_balloon_instance \                      
    --img-size 640 640 \                                                                                     
    --batch 1 \                                                                                              
    --device cpu \                                                                                           
    --simplify \                                                                                             
        --opset 11 \                                                                                         
        --pre-topk 1000 \                                                                                    
        --keep-topk 100 \                                                                                    
        --iou-threshold 0.65 \                                                                               
        --score-threshold 0.25                                                                               

However, I get the following error:

Export ONNX with bbox decoder and NMS ...                                                                                                       
!!!You are using `YOLOv5Head` with num_classes == 1. The loss_cls will be 0. This is a normal phenomenon.                                       
Loads checkpoint by local backend from path: work_dirs/yolov5_ins_s-v61_syncbn_fast_8xb16-300e_balloon_instance/best_coco_bbox_mAP_epoch_240.pth
Traceback (most recent call last):                                                                                                              
  File "projects/easydeploy/tools/export_onnx.py", line 157, in <module>                                                                        
    main()                                                                                                                                      
  File "projects/easydeploy/tools/export_onnx.py", line 116, in main                                                                            
    deploy_model(fake_input)
  File "/home/motiv/anaconda3/envs/mmyolo/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl                     
    return forward_call(*input, **kwargs)
  File "/home/motiv/nkodenko/mmyolo/mmyolo/projects/easydeploy/model/model.py", line 166, in forward
    return self.pred_by_feat(*neck_outputs)
TypeError: pred_by_feat() takes from 3 to 4 positional arguments but 6 were given

Is export to onnx supported for instance segmentaiton toasks in yolo?

[BaderTim]

@nik123 did you manage to solve this issue? I am running to the same error