Detectron2 is FAIR's next-generation platform for object detection, segmentation and other visual recognition tasks.
It is essential that you have a detectron2 environment:
See installation instructions.
Please run the following command to have a softlink to the best model weights, where your directory is the folder in which you cloned the repository:
ln -s /media/user_home1/idchacon/detectron2/projects/DensePose/exp4_best_model your_directory/detectron2/projects/DensePose
You will also need a softlink to the dataset:
ln -s /media/user_home1/idchacon/detectron2/datasets your_directory/detectron2/projects/DensePose
You will find the main.py by running the following command:
cd projects/Densepose
To replicate the results of the best model described in the paper you can run the following command:
python main.py --mode test
To visualize the performance of the model over an image you can run the following command:
python main.py --mode demo --img image_0.png
where image_0.png is the image that will be tested. By default, you will encounter the image cicla.jpg.
After this process, you will obtain a series of images according to the U_coordinates, V_coordinates, Segmentation and Contour_plots.