PyQT5 based GUI application for performance comparison between Tensorflow (CPU) and PlaidML (GPU)
CPU : Intel i7-8086K
GPU : Gigabyte Radeon RX Vega 64 Silver HBM2 8GB
OS : Ubuntu 18.04
Driver : AMD-GPUPRO 18.20
Python 3.5
Keras 2.2.2
Tensorflow 1.10
PlaidML 0.3.5
We used SSD (single shot multibox detector) model for internal object detection algorithm.
Also, we were able to develop this application with a lot of inspiration from the repository below.
Download the pre-trained weights from ssd_kerasV2 named VGG16SSD300weights_voc_2007_class20.hdf5
Save the weight file in the weight
directory.
Create anaconda environment using environment.yml
and following command:
$ conda env create
Activate virtual envitonment and start the application
$ conda activate demo
(demo) $ python main.py