Skip to content

README.md

Latest commit

 

History

History
64 lines (55 loc) · 1.63 KB

README.md

File metadata and controls

64 lines (55 loc) · 1.63 KB
Raw

Instructions for OminiControl

Installation

Move to the OminiControl folder:

cd OminiControl

Then run:

uv sync --index-strategy unsafe-best-match

Training

Replace <config_file> withe the correct config file in train/config folder and run:

export OMINI_CONFIG=./train/config/<config_file>
uv run accelerate launch --main_process_port 41353 -m omini.train_flux.train_spatial_alignment_v2

Evaluation

To generate and evaluate the control models use the followings

Depth

uv run python generate_depth_eval_samples.py \
  --ckpt_path runs/<path_to_ckpt_file> \
  --data_path <path_to_depth_data_path> \
  --image_size 512 \
  --save_path <save_path>

Semantic Segmentation

uv run generate_seg_eval_samples.py \
  --ckpt_path runs/<path_to_ckpt_file> \
  --input_csv <path_to_parent_directory>/ControlNet/ade20k_val_caption.csv \
  --image_size 512 \
  --save_path <save_path> \
  --seed 123

Canny Edge

uv run generate_canny_eval_samples.py \
  --ckpt_path runs/<path_to_ckpt_file> \
  --input_csv <path_to_parent_directory>/ControlNet/ade20k_val_caption.csv \
  --image_size 512 \
  --save_path <save_path> \
  --seed 123

Pose

uv run generate_pose_eval_samples.py \
  --ckpt_path runs/<path_to_ckpt_file> \
  --data_path <path_to_parent_directory>/ControlNet/control_datasets/coco_keypoints/val \
  --metadata_path <path_to_parent_directory>/ControlNet/control_datasets/coco_keypoints/val/metadata.csv \
  --keypoints_json <path_to_parent_directory>/ControlNet/person_keypoints_val2017.json \
  --image_size 512 \
  --save_path <save_path> \
  --seed 123