Closed VincentDuf closed 4 years ago
Hey, can you change entry_points = {'console_scripts': ['retinanet=retinanet.main:main']}
to entry_points = {'console_scripts': ['odtk=retinanet.main:main']}
Also try extending libraries to libraries=['nvinfer', 'nvinfer_plugin', 'nvonnxparser', 'opencv_core', 'opencv_highgui', 'opencv_imgproc', 'opencv_imgcodecs'])
and
library_dirs to ['/usr/local/lib', '/usr/local/lib64/']
Just noticed I forgot the _"opencvimgcodecs" in the library list... it works now.
Please see here for MobileNetV2FPN pre-trained model: https://github.com/NVIDIA/retinanet-examples/releases/tag/v0.2.3
Please see here for MobileNetV2FPN pre-trained model: https://github.com/NVIDIA/retinanet-examples/releases/tag/v0.2.3
this link can not open
Hi guys, I've been working with the repo for a few weeks now and everything went great until I tried to update the repo to get the new MobileNet implementation.
I tried to rebuild the repo after I cloned, everything seems to go well (I can even see the odtk in pip list) but when I invoke it I have the "undefined symbol" error.
I'm working with an Ubuntu16.04 host, CUDA10.2, TensorRT7, Tensorflow1.15.
I want to precise that I made a second "setup.py" script with modifications:
With this I get "retinanet" to be installed (because of the change in certain variable) but I get the same error when I invoke the command...
If you have any ideas how to deal with it I take ! It's even more frustrating that I had the issue when I first attempt to compile the repo and by adding the changes it worked, I don't get why it doesn't now.
Thanks