Object detection slow and does not use GPU

Question

I need to use Tensorflow Object Detection API to make some classification connected with recognition.

My problem is that using the API for detection with a pretrained coco model takes too much time and for sure does not use the GPU. I checked my tensorflow-gpu installation on different scripts and it works fine, but when I use this model for detection I can only see increse in CPU usage.

I checked different version of tensorflow (1.12, 1.14), different combinations of CUDA Toolkit (9.0, 10.0) and CuDNN (7.4.2, 7.5.1, 7.6.1) but it is all the same, also tried it on both Windows 7 and Ubuntu 16.04, no difference. My project however requires much faster detection time.

System information: System: Windows 7, Ubuntu 16.04 Tensorflow: 1.12, 1.14 GPU: GTX 970

Pranit Kothari · Accepted Answer · 2019-07-28 10:56:13Z

2

Run following python code, if it detects GPU then you can use GPU for training otherwise there is some problem,

from tensorflow.python.client import device_lib
print(device_lib.list_local_devices())

One more thing, just because your CPU is utilizing does not mean GPU is not at work. CPU always be busy, GPU should also spike when you are training.

Paste the output of above code in the comment if you are not sure about the output.

Edit: After chat with OP on comments, I see the suggested code, and it is using pretrained model, so no training happening here. You are using model and not training a new model. So no gpu is being used.

edited Jul 28, 2019 at 10:56

answered Jul 28, 2019 at 9:53

Pranit Kothari

9,92511 gold badges65 silver badges145 bronze badges

Sign up to request clarification or add additional context in comments.

9 Comments

GKozinski Over a year ago

Thanks for your answer. [name: "/device:CPU:0" device_type: "CPU" memory_limit: 268435456 locality { } incarnation: 12868789687935921631 , name: "/device:GPU:0" device_type: "GPU" memory_limit: 3546435584 locality { bus_id: 1 links { } } incarnation: 5695922347637991756 physical_device_desc: "device: 0, name: GeForce GTX 970, pci bus id: 0000:01:00.0, compute capability: 5.2" ]

GKozinski Over a year ago

I also checked the usage of GPU with GPU-Z software for monitoring GPU work and I did not notice even a small spike or VRAM utilisation.

Pranit Kothari Over a year ago

Yes, the output is correct, it is suggesting you can use GPU, not sure what is wrong if possible can you paste your code, also try using google colab if you still face same issue there.

GKozinski Over a year ago

Right now I only use the exemplary code from delivered with the library Jupyter Notebook object_detection_tutorial.ipynb. I just wanted to use it to check how long does it take to make the detections, since it is crucial for my project. I see some people on the internet report something like even 10 detections per second, when I get 1 detection in 5 seconds.

Pranit Kothari Over a year ago

@Makintosz Ah!, you are not training you are using pretrained model. Look at my answer. I have edited it, you are not training model here so no GPU will be required. Please accept the answer, if I have solved your problem.

|

Collectives™ on Stack Overflow

Object detection slow and does not use GPU

1 Answer 1

9 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

9 Comments

Your Answer

Sign up or log in

Post as a guest

Related