有没有人把这个东西运行在 NVIDIA 的JETSON NANO中的 ，求教程

[albertyou2]

有没有人把这个东西运行在 NVIDIA 的JETSON NANO中的 ，求教程 谢谢！

[c4b4d4]

I'm trying to run this on Jeston Nano but I get stuck on this:

out_boxes, out_scores, out_classes = self.sess.run(
           [self.boxes, self.scores, self.classes],
           feed_dict={
               self.yolo_model.input: image_data,
               self.input_image_shape: [image.size[1], image.size[0]],
               K.learning_phase(): 0
           })
       return_boxs = []

It hangs out at this level forever. What could it be causing it?

[c4b4d4]

After leaving it for a while, it displays the following message:

kthreadd: page allocation stalls for 10004ms, order:2, mode:0x27080c0(GFP_KERNAL_ACCOUNT|__GFP_ZERO|__GFP_NOTRACK)
kthreadd invoked oom-killer: gfp_mask=0x27080c0(GFP_KERNEL_ACCOUNT|__GFP_ZERO|__GFP_NOTRACK), nodemask=0, order=2, oom_score_adj=0
Out of memory: Kill process 7248 (python3) score 266 or sacrifice child

[albertyou2]

看上去这内存分配失败了，是不是内存/显存不足？

[c4b4d4]

Memory is hitting the limit, just verified it with tegrastats

I'm adding a swapfile to help the the memory, will let you know if that fixed it.

[albertyou2]

@kzka90 Thank you very much !

[huachao2017]

tensorflow part in generate_detections: config = tf.ConfigProto() config.gpu_options.per_process_gpu_memory_fraction = 0.1 # GPU10% self.session = tf.Session(config=config)

keras part in yolo: config = tf.ConfigProto() config.gpu_options.per_process_gpu_memory_fraction = 0.2 # GPU20% set_session(tf.Session(config=config))

[linspace100]

@kzka90 did you solve the issue with a swapfile? I made a 6GB swapfile. But I can`t do well on my nano.

[JanEveraertEHB]

this seems to be running stuck on the following issue for me:

tensorflow.python.framework.errors_impl.ResourceExhaustedError: OOM when allocating tensor with shape[3,3,512,1024] and type float on /job:localhost/replica:0/task:0/device:GPU:0 by allocator GPU_0_bfc [Op:Mul]

8GB allocated for memory, should a swap file help?

Also: would this work in a live setup, pulling images from the camera?