Steps for generating data

go to data_generator folder
run objects_detection_v3.py to get bboxes of objects in ImageNet images
run mask_imgnet.py to mask ImageNet images
(Optional) run data_augmentation.py to wrap foreground objects with garment images
(Optional) run data_pair_preparation.py to generate training data pairs list in data_pairs.txt

Outputs structure

all_bboxes.json: bboxes of objects in a image, {"img_name.JPEG": [bbox1, bbox2,...]}
foreground: masked objects, "img_namemaski_foreground.png"
inpaint: masked background, "img_namemaski_background.png"
model: original image, "img_name.JPEG"
mask: mask image, "img_namemaski_mask.png"
visualization: visualization of bounding boxes, "annoated_img_name.png"

Name		Name	Last commit message	Last commit date
Latest commit History < 8000 svg aria-hidden="true" focusable="false" class="octicon octicon-history" viewBox="0 0 16 16" width="16" height="16" fill="currentColor" display="inline-block" overflow="visible" style="vertical-align:text-bottom">11 Commits
.vscode		.vscode
data_generator		data_generator
.gitignore		.gitignore
README.MD		README.MD
count_generated_imgs.py		count_generated_imgs.py
environment.yml		environment.yml
install_sam2.sh		install_sam2.sh
utils.py		utils.py