pangchao-git

硬件平台: Jesson xavier NX etPack4.4；cuDNN 8.0；CUDA 10.2；OpenCV 4.1.1；Visionworks 1.6；python3 3.6.9 ；TensorRT 7.1.3.4 修改config Config config_v3; config_v3.net_type = YOLOV3; config_v3.file_model_cfg = "../configs/yolov3.cfg"; config_v3.file_model_weights = "../configs/yolov3-voc_11000.weights"; config_v3.calibration_image_list_file_txt = "../configs/calibration_images.txt"; config_v3.inference_precison =INT8; config_v3.detect_thresh...

mmdetection maskrcnn to deepstream, 如何支持在deepstream获取mask

1

1 模型转化时候这样修改了Infrence.py trt_model = mmdet2trt(cfg_path, args.checkpoint, output_names=["num_detections", "boxes", "scores", "classes","masks"], fp16_mode=args.fp16, device=args.device,enable_mask=True) torch.save(trt_model.state_dict(), args.save_path) 2 deepstream 的配置文件 output-bbox-name=bbox output-blob-names=num_detections;boxes;scores;classes;masks parse-bbox-func-name=NvDsInferParseMmdet custom-lib-path=nvdsinfer_custom_impl_fasterRCNN/libamirstan_plugin.so

1 模型转化时候这样修改了Infrence.py trt_model = mmdet2trt(cfg_path, args.checkpoint, output_names=["num_detections", "boxes", "scores", "classes","masks"], fp16_mode=args.fp16, device=args.device,enable_mask=True) torch.save(trt_model.state_dict(), args.save_path) 2 deepstream 的配置文件 output-bbox-name=bbox output-blob-names=num_detections;boxes;scores;classes;masks parse-bbox-func-name=NvDsInferParseMmdet custom-lib-path=nvdsinfer_custom_impl_fasterRCNN/libamirstan_plugin.so

运行后出现void check_error(cudaError_t): Assertion `0' failed. CUDA Error: no kernel image is available for execution on the device

1

大佬，编译通过后，运行时候出现void check_error(cudaError_t): Assertion `0' failed. CUDA Error: no kernel image is available for execution on the device，这个问题需要怎么解决？

目标硬件平台:rk3588 rknn-tookit版本:rknn-toolkit2-1.3.0 mmdetection模型框架: faster fpn 101 参考onnx转rknn的下的resnet50v2/test.py去做模型build 和export的 未开启量化时候报错  开启时候报错  请rk开发人员看看是啥问题