FreeCP: Training-Free Class Purification for Open-Vocabulary Semantic Segmentation

Qi Chen¹²,   Lingxiao Yang¹,   Yun Chen³,   Nailong Zhao⁴,   Jianhuang Lai¹,   Jie Shao²,   Xiaohua Xie¹,  

¹Sun Yat-sen University,   ²ByteDance Intelligent Creation,   ³University of Surrey,   ⁴Alibaba Cloud Computing

• [Paper] •

🚀 Overview

📦 Dependencies and Installation

# git clone this repository
git clone https://github.com/chenqi1126/FreeCP.git
cd FreeCP

# create new anaconda env
conda create -n freecp python=3.10
conda activate freecp

# install torch and dependencies
pip install -r requirements.txt

🗂️ Datasets

Please follow the MMSeg data preparation document and ClearCLIP to download and pre-process the datasets.

dataset
├── ADE
│   ├── ADEChallengeData2016
│   │   ├── annotations
│   │   │   ├── training
│   │   │   ├── validation
│   │   ├── images
│   │   │   ├── training
│   │   │   ├── validation
├── Cityscapes
│   ├── leftImg8bit
│   │   ├── train
│   │   ├── val
│   ├── gtFine
│   │   ├── train
│   │   ├── val
├── ms_coco_17
│   ├── images
│   │   ├── train2017
│   │   ├── val2017
│   ├── annotations
│   │   ├── object
│   │   ├── stuff
├── PascalVOC
│   ├── VOC2012
│   │   ├── JPEGImages
│   │   ├── SegmentationClass
│   │   ├── ImageSets
│   │   │   ├── Segmentation
│   ├── VOC2010
│   │   ├── JPEGImages
│   │   ├── SegmentationClassContext
│   │   ├── ImageSets
│   │   │   ├── SegmentationContext
│   │   │   │   ├── train.txt
│   │   │   │   ├── val.txt

🚀 Model evaluation

Single-GPU:

python eval.py --config ./config/cfg_DATASET.py --workdir YOUR_WORK_DIR

Multi-GPU:

bash ./dist_test.sh ./config/cfg_DATASET.py

📈 Results

Method	VOC21	PC60	Object	VOC20	Cityscapes	PC59	ADE	Stuff	Average
MaskCLIP	43.4	23.2	20.6	74.9	24.9	26.4	11.9	16.7	30.3
+ FreeCP	64.4	34.7	36.2	84.1	32.5	36.6	17.6	23.3	41.2
GEM	56.9	32.6	31.1	79.9	30.8	35.9	15.7	23.7	38.3
+ FreeCP	64.7	35.5	36.9	80.6	35.7	39.1	17.8	25.8	42.0
SCLIP	59.1	30.4	30.5	80.4	32.2	34.2	16.1	22.4	38.2
+ FreeCP	65.8	35.3	37.2	84.3	33.3	38.0	18.4	24.9	42.1
ClearCLIP	51.8	32.6	33.0	80.9	30.0	35.9	16.7	23.9	38.1
+ FreeCP	64.5	35.7	36.9	81.5	34.4	39.3	18.9	26.1	42.2

📚 Citation

@inproceedings{chen2025freecp,
  title={Training-Free Class Purification for Open-Vocabulary Semantic Segmentation},
  author={Qi Chen, Lingxiao Yang, Yun Chen, Nailong Zhao, Jianhuang Lai, Jie Shao, Xiaohua Xie},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  year={2025}
}

🙏 Acknowledgement

This implementation is based on ClearCLIP, SCLIP and CLIP-ES. Thanks for the awesome work.