zengwb-lx

Results 20 issues of zengwb-lx

1850行处:const int MAX_IMAGE_BBOX = 1024 1024设的太小了,会出现漏检的情况,应该开放出来或设大点

风格迁移onnx模型8.9mb,转成rknn(toolkit2是1.5.2版本)后变成了429mb,不量化和onnx输出可对齐,量化后输出差异很大,qq群回复是npu任务指令异常,附上onnx模型,以便排查验证 onnx模型: [face_animev2.zip](https://github.com/rockchip-linux/rknn-toolkit2/files/13657134/face_animev2.zip)

你好, 请问你开源出来的模型性能怎么样?是如图casia_phase2.jpg里的最好模型吗? 还有一问题就是:模型用depth图推理时输出的depth_prob,该设多大阈值确定真假, 还是说必须1 is real;0 is fake,谢谢

hello, can you tell me what 'data/tmp/pic_path' is .thanks! print("=> Start copy pictures to the output path {} ......".format(args.output_picture_path)) with open('data/tmp/pic_path', 'r') as f: content = f.read() picture_path_dict = json.loads(content)

I'm new to Openvino, can you provide the CMakeLists.txt. thx!

请问有没有yolov5-seg的支持呢,谢谢

你好,我用VS2015编译,在CUDA代码cuda_function.cu报了关于核函数的错 ![image](https://github.com/Li-Hongda/TensorRT_Inference_Demo/assets/52445450/226c74b8-405a-4368-a083-07abb4f47576) 请问有支持windows的计划吗

how to train yolo-seg

Is there yolo-seg implementation? thanks

question