Labelled Artificially Occluded Panda Data as Training Input for Attention-UNet GAN Inpainting Model
收藏资源简介：
This dataset contains aritificially occluded images in images.zip and corresponding ground truth labels that is unoccluded images of the Franka Emika Panda arm are in labels.zip Once downloaded follow these steps to create the final re-useable dataset: Create a new folder say <Attention-UNet Inpainting Dataset> Extract images.zip in a folder named 'images' Extract labels.zip in a folder named 'labels' Copy both 'images' and 'labels' folders inside <Attention-UNet Inpainting Dataset> <Attention-UNet Inpainting Dataset> after populating is used to train the novel Attention-UNet GAN Inpainting model for occlusion handling The details on how to use this dataset is detailed in the following Repo: README.md with instructions on building the occlusion inpainting model training network
