search

for-aiur / yolo_cpp

C++ed version of Yolo
Other
69 stars 37 forks source link

Detection failed and got errors #5

Open zlt1991 opened 6 years ago

zlt1991 commented 6 years ago

Hi, it is a wonderful work!. I'm freshman to the yolo detector, and I encountered some problems while simply using this yolo c++ wrapper.

I complied the source code and no error occurred. Then put my custom data near the darknet, which included .cfg, weights, .name, .data. However, when I run following command:
./darknet detector test cfg_zlt/zlt.data cfg_zlt/yolo-zlt.cfg backup/yolo-zlt_900.weights data/nfpa/pos-123.jpg

No detection result outputted. It seemed the darknet didn't detect anything. But in fact, I got a correct detection results when used the darcknet compiled from original code.

In addition, I tried to train my custom dataset using your darknet, and got strange outputs like this:

memset begins memset ends for bbbb for jjjj for bbbbb for jjjj for bbbbb for jjjj for bbbbb for jjjj for bbbbb for jjjj for bbbbb for jjjj for bbbbb for jjjj for bbbbb for jjjj for bbbbb for finished Region Avg IOU: 0.171883, Class: 1.000000, Obj: 0.001114, No Obj: 0.003235, Avg Recall: 0.250000, count: 8

It crashed after several epochs and the error message was given: CUDA Error: invalid configuration argument darknet: /home/zlt/packages/yolo_cpp/src/cuda.cpp:36: void check_error(cudaError_t): Assertion `0' failed. Aborted (core dumped)

Any suggestions?

achalshah20 commented 6 years ago

I am also facing the same issue. Any update?

wangjhit commented 6 years ago

The same error occurred when I trained my dataset. Is there a solution now?