CODEWITHSUNDEEP
segmentation-faultcross-compilingnao-robotdarknet
Eino Gourdin
Eino Gourdin

Reputation: 4537

Segmentation fault when running darknet classifier on Nao

I'm trying to run darknet imagenet classifier on Nao, but it crashes with a segfault.

With the YOLO config (./darknet detect cfg/yolo.cfg yolo.weights data/dog.jpg), darknet runs, but trying to run classifier (./darknet classifier predict cfg/imagenet1k.data cfg/extraction.cfg extraction.weights data/dog.jpg) only produces a segfault :

$ ./darknet  classifier predict cfg/imagenet1k.data cfg/extraction.cfg extraction.weights data/dog.jpg
layer     filters    size              input                output
    0 conv     64  7 x 7 / 2   224 x 224 x   3   ->   112 x 112 x  64
    1 max          2 x 2 / 2   112 x 112 x  64   ->    56 x  56 x  64
    2 conv    192  3 x 3 / 1    56 x  56 x  64   ->    56 x  56 x 192
    3 max          2 x 2 / 2    56 x  56 x 192   ->    28 x  28 x 192
    4 conv    128  1 x 1 / 1    28 x  28 x 192   ->    28 x  28 x 128
    5 conv    256  3 x 3 / 1    28 x  28 x 128   ->    28 x  28 x 256
    6 conv    256  1 x 1 / 1    28 x  28 x 256   ->    28 x  28 x 256
    7 conv    512  3 x 3 / 1    28 x  28 x 256   ->    28 x  28 x 512
    8 max          2 x 2 / 2    28 x  28 x 512   ->    14 x  14 x 512
    9 conv    256  1 x 1 / 1    14 x  14 x 512   ->    14 x  14 x 256
   10 conv    512  3 x 3 / 1    14 x  14 x 256   ->    14 x  14 x 512
   11 conv    256  1 x 1 / 1    14 x  14 x 512   ->    14 x  14 x 256
   12 conv    512  3 x 3 / 1    14 x  14 x 256   ->    14 x  14 x 512
   13 conv    256  1 x 1 / 1    14 x  14 x 512   ->    14 x  14 x 256
   14 conv    512  3 x 3 / 1    14 x  14 x 256   ->    14 x  14 x 512
   15 conv    256  1 x 1 / 1    14 x  14 x 512   ->    14 x  14 x 256
   16 Segmentation fault (core dumped)

Core dump is not available, as /proc/sys/kernel/core_pattern contains only |/bin/false

But running it with gdb, I could get the crash stack : 

#0  0x0806efac in make_convolutional_layer ()
#1  0x080a4919 in parse_convolutional ()
#2  0x080a6e11 in parse_network_cfg ()
#3  0x0805d7ef in predict_classifier ()
#4  0x0805e85c in run_classifier ()
#5  0x080499c0 in main ()

I see make_convolutional_layer allocates a bunch of memory. Could the crash be the program reaching memory limit ? However In YOLO mode, it builds a bigger network (with greater layers size), so it doesn't sound too logical. Any idea ?

Upvotes: 1

Views: 1449

Answers (1)

Eino Gourdin
Eino Gourdin

Reputation: 4537

Actually it's caused by a memory shortage, a call to calloc that returns null. (and that seems to always happen on the line l.weights = calloc(c*n*size*size, sizeof(float)); in make_convolutional_layer, that on the 16th layer tries to allocate 4 718 592 bytes)

So there doesn't seem to be an answer to the problem, apart from trying to build a smaller network or increase available memory.

Edit: The smallest "Darknet Reference" network runs, the others are too heavy for Nao.

Upvotes: 1

Related Questions