0why0/S2RL

⚗️ S²-RL

Sample-Set Dual Reinforcement Learning
for Generative Semantic Segmentation Dataset Distillation

Haoyu Wang · Fei Zhou · Qingqing Qiu · Lei Zhang · Wei Wei · Chen Ding

📄 arXiv · 🌐 Project Page · 💻 Code · 📂 Datasets · 🚀 Policy Training · 🎨 Generation

This repository contains the ADE20K and COCO datasets and LoRA adapters used in the S²-RL main experiments: 10 data archives, 21,700 image–mask pairs, and two dataset-specific adapters.

📂 Datasets

Selected real-data subsets

The selection baselines provide subsets of real images and their ground-truth masks. Uniform subsets also supply the category prompts for S²-RL.

Method	ADE20K	COCO
Random	Download	Download
Uniform	Download	Download
Herding	Download	Download
K-Center	Download	Download

All data archives include three budgets and five sets per budget, with paired JPEG images and PNG segmentation masks.

Dataset	Ratios	Image–mask pairs per set
ADE20K	0.0025, 0.005, 0.01	50, 101, 202
COCO	0.000625, 0.00125, 0.0025	73, 147, 295

Extract archives into the data root used by the code:

tar -xzf <archive>.tar.gz -C "$S2RL_DATA_ROOT"

ILDistillation/<dataset>/<method>/
├── images/time_<n>_ratio_<r>/*.jpg
├── annotations/time_<n>_ratio_<r>/*.png
└── _manifest.json

ADE20K masks use labels 1..150, with 0 ignored. COCO masks use 0..132, with 255 ignored; class order follows dataset/coco.py.

🚀 Policy Training

Policy training and generation are described in the code repository.

Dataset	Policy weights	Adapter config
ADE20K	Download	Download
COCO	Download	Download

These LoRA adapters use Stable Diffusion 3.5 Medium as the base model. Save both files in the same directory and set POLICY_DIR in the generation script to that directory.

🎨 Generation and selection

Distilled datasets

S²-RL provides distilled datasets of generated images and pseudo-labels for downstream segmentation training.

Method	ADE20K	COCO
S²-RL	Download	Download

Extract into $S2RL_DATA_ROOT. Image–mask pairs are stored under ILDistillation/<dataset>/S2RL/; use DATASET_NAME="S2RL" in the downstream evaluation scripts.

🖊️ Citation

@misc{wang2026s2rl,
  title={S$^2$-RL: Sample-Set Dual Reinforcement Learning for Generative Semantic Segmentation Dataset Distillation},
  author={Haoyu Wang and Fei Zhou and Qingqing Qiu and Lei Zhang and Wei Wei and Chen Ding},
  year={2026},
  eprint={2610.35491},
  archivePrefix={arXiv},
  primaryClass={cs.CV},
  url={https://arxiv.org/abs/2610.35491}
}

📄 License

S²-RL contributions are licensed under the MIT License. Third-party components and assets follow their original licenses.