Code release for "Cut and Learn for Unsupervised Object Detection and Instance Segmentation" and "VideoCutLER: Surprisingly Simple Unsupervised Video Instance Segmentation"
Hi so i was using cutler and decided to convert into trt engine to enhance the efficiency,
So as you know your pretrained cutler model is based on detectron2 framework and they have a step by step process to convert a model to onnx first and then to trt.
While i was able to create onnx file using first conversion ,
during secondary conversion i am getting issues
File "TensorRT/samples/python/detectron2/create_onnx.py", line 482, in roi_heads bbox_pred_reshape = self.graph.op_with_const("Reshape", "bbox_pred/reshape", bbox_pred.outputs[0], bbox_pred_shape) AttributeError: 'NoneType' object has no attribute 'outputs'
Do you guys have any support regarding this on how to convert model into appropriate trt convertable onnx file for Custom roi heads/Culter?
Hi so i was using cutler and decided to convert into trt engine to enhance the efficiency,
So as you know your pretrained cutler model is based on detectron2 framework and they have a step by step process to convert a model to onnx first and then to trt.
While i was able to create onnx file using first conversion , during secondary conversion i am getting issues
File "TensorRT/samples/python/detectron2/create_onnx.py", line 482, in roi_heads bbox_pred_reshape = self.graph.op_with_const("Reshape", "bbox_pred/reshape", bbox_pred.outputs[0], bbox_pred_shape) AttributeError: 'NoneType' object has no attribute 'outputs'
Do you guys have any support regarding this on how to convert model into appropriate trt convertable onnx file for Custom roi heads/Culter?