search

pjreddie / darknet

Convolutional Neural Networks
http://pjreddie.com/darknet/
Other
25.73k stars 21.33k forks source link

Segmentation fault when train tiny-yolo #178

Open Sander-houqi opened 7 years ago

Sander-houqi commented 7 years ago

Hi:

when I use ./darknet detector train cfg/voc.data cfg/tiny-yolo-voc.cfg tiny-yolo-voc.weights to train my own datasets,it cause Segmentation fault,but I dont't how to fix it .

here is gdb results by use command: gdb --args ./darknet detector train cfg/voc.data cfg/tiny-yolo-voc.cfg tiny-yolo-voc.weights

[New Thread 0x7fffe7887700 (LWP 18472)] [New Thread 0x7fffe9719700 (LWP 18473)] [New Thread 0x7fffe9f1a700 (LWP 18474)] [New Thread 0x7fffeaf1c700 (LWP 18475)] [New Thread 0x7fffea71b700 (LWP 18476)] [New Thread 0x7fffe7086700 (LWP 18477)] [New Thread 0x7fffd6de7700 (LWP 18478)] [New Thread 0x7fffd65e6700 (LWP 18479)] Region Avg IOU: -nan, Class: -nan, Obj: -nan, No Obj: 0.000426, Avg Recall: -nan, count: 0 Region Avg IOU: -nan, Class: -nan, Obj: -nan, No Obj: 0.000469, Avg Recall: -nan, count: 0 Region Avg IOU: -nan, Class: -nan, Obj: -nan, No Obj: 0.000421, Avg Recall: -nan, count: 0 Region Avg IOU: -nan, Class: -nan, Obj: -nan, No Obj: 0.000417, Avg Recall: -nan, count: 0 Region Avg IOU: -nan, Class: -nan, Obj: -nan, No Obj: 0.000415, Avg Recall: -nan, count: 0 Region Avg IOU: 0.605891, Class: 1.000000, Obj: 0.000000, No Obj: 0.000424, Avg Recall: 1.000000, count: 1 Region Avg IOU: 0.519688, Class: 1.000000, Obj: 0.000000, No Obj: 0.000462, Avg Recall: 1.000000, count: 1

Program received signal SIGSEGV, Segmentation fault. 0x000000000046ac60 in get_region_box (x=0xc84b440, biases=0x8d1400, n=0, index=-2014970546, i=18999980, j=18999980, w=19, h=19, stride=361) at ./src/region_layer.c:79 79 b.x = (i + x[index + 0*stride]) / w;

(gdb) where

0 0x000000000046ac60 in get_region_box (x=0xc84b440, biases=0x8d1400, n=0, index=-2014970546, i=18999980, j=18999980,

w=19, h=19, stride=361) at ./src/region_layer.c:79

1 0x000000000046bf8b in forward_region_layer (l=..., net=...) at ./src/region_layer.c:274

2 0x000000000046d6e4 in forward_region_layer_gpu (l=..., net=...) at ./src/region_layer.c:556

3 0x000000000047bc99 in forward_network_gpu (net=...) at ./src/network_kernels.cu:47

4 0x000000000047c193 in train_network_datum_gpu (net=...) at ./src/network_kernels.cu:126

5 0x0000000000452312 in train_network_datum (net=...) at ./src/network.c:264

6 0x00000000004525e8 in train_network (net=..., d=...) at ./src/network.c:299

7 0x000000000041d6e6 in train_detector (datacfg=0x7fffffffe498 "cfg/voc.data",

cfgfile=0x7fffffffe4a5 "cfg/tiny-yolo-voc.cfg", weightfile=0x7fffffffe4bb "tiny-yolo-voc.weights", 
gpus=0x7fffffffdfa4, ngpus=1, clear=0) at ./examples/detector.c:116

8 0x000000000042126d in run_detector (argc=6, argv=0x7fffffffe198) at ./examples/detector.c:698

9 0x0000000000426165 in main (argc=6, argv=0x7fffffffe198) at ./examples/darknet.c:438

I don't know why index is -2014970546? I guess that is out of range ,but I can't find the resolutions,can anybody help me ,thanks！

jremillard commented 7 years ago

See issue #122