Views
No views yet
Sa2va-Instance-4B checkpoint, which serves as Stage 1 (Instance Parsing Model) for InstanceControl, presented in the paper InstanceControl: Controllable Complex Image Generation without Instance Labeling.Sa2va-Instance-4B model—to automatically parse instance descriptions from text prompts and predict instance masks based on visual conditions (such as Canny edges, depth, or HED).1python stage1_Sa2VA/projects/llava_sam2/evaluation/gcg_eval_our_folders.py \
2 --model_path /path/to/Sa2va-Instance-4B \
3 --image_dir ./example/canny \
4 --json_dir ./example/json \
5 --save_dir ./results/json_pred_canny1@article{liu2026instancecontrol,
2 title={InstanceControl: Controllable Complex Image Generation without Instance Labeling},
3 author={Xiaoyu Liu and Huan Wang and Fan Li and Zhixin Wang and Jiaqi Xu and Ming Liu and Wangmeng Zuo},
4 journal={arXiv preprint arXiv:2606.31924},
5 year={2026}
6}