Scaled-YOLOv4-TensorRT icon indicating copy to clipboard operation
Scaled-YOLOv4-TensorRT copied to clipboard

Published 20 hours ago • verified publisher icon tjuskyzhang

kernel weights has count 9216 but 87040 was expected

Open mate-hegedus opened this issue 3 years ago • 0 comments

Please help me out with this one: I try to convert sample yolov4-p6.pt.. after generating wts i receive this:

"root@3bc22b0a0ace:/yolo/Scaled-YOLOv4-TensorRT/yolov4-p6-tensorrt/build# ./yolov4-p6 -s 1.0 1.0 Loading weights: ../yolov4-p6.wts Building engine, please wait for a while... [03/09/2022-19:00:09] [E] [TRT] (Unnamed Layer* 665) [Convolution]: kernel weights has count 9216 but 87040 was expected [03/09/2022-19:00:09] [E] [TRT] (Unnamed Layer* 665) [Convolution]: count of 9216 weights in kernel, but kernel dimensions (1,1) with 256 input channels, 340 output channels and 1 groups were specified. Expected Weights count is 256 * 11 * 340 / 1 = 87040 [03/09/2022-19:00:09] [E] [TRT] Could not compute dimensions for (Unnamed Layer 665) [Convolution]_output, because the network is not valid. [03/09/2022-19:00:09] [E] [TRT] Network validation failed. Build engine successfully! yolov4-p6: /yolo/Scaled-YOLOv4-TensorRT/yolov4-p6-tensorrt/yolov4-p6.cpp:143: void APIToModel(unsigned int, nvinfer1::IHostMemory**, float&, float&): Assertion `engine != nullptr' failed. Aborted (core dumped)"

mate-hegedus avatar Mar 09 '22 19:03 mate-hegedus