RAM

KO | EN

gitlite — search

#artificial-intelligence #mysql #github #ai #database #postgresql #mcp #animation #csharp #linux

(★ 29)

[CVPR2025] Official implementation of RAM

File Explorer

ZIP Download

- __init__.py
- adapters.py
- bpe_simple_vocab_16e6.txt.gz
- clip.py
- model.py
- simple_tokenizer.py
- __init__.py
- defaults.py
- coco.yml
- nus.yml
- - test_1006_filtered.json
  - test_17_filtered.json
  - test_65_filtered.json
  - test_81_filtered.json
  - train_48_filtered.json
  - train_81_filtered.json
- - classification.py
- __init__.py
- bases.py
- make_dataloader.py
- __init__.py
- asymmetric_loss.py
- mmc_loss.py
- seesawloss.py
- __init__.py
- base.py
- model.py
- ot_solver.py
- __init__.py
- processor.py
- __init__.py
- make_optimizer.py
- make_scheduler.py
- method.png
- __init__.py
- logger.py
- meter.py
- metrics.py
- model_utils.py
.gitignore
README.md
requirements.txt
train.py

# Use via CDN

jsDelivr

jsDelivr serves any public GitHub repository as a CDN with zero setup. Pick a version and a file to get a ready-to-paste link and snippet.

Version

File

Link

Example

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

Official implementation of the [paper](https://arxiv.org/abs/2503.15337) in CVPR 2025:

**Recover and Match: Open-Vocabulary Multi-Label Recognition through Knowledge-Constrained Optimal Transport**

[Hao Tan](https://scholar.google.com/citations?hl=zh-CN&user=gPEjNFcAAAAJ), [Zichang Tan](https://scholar.google.com/citations?user=s29CDY8AAAAJ&hl=zh-CN&oi=ao), [Jun Li](https://bazinga699.github.io/), [Ajian Liu](https://scholar.google.com/citations?user=isWtY64AAAAJ&hl=zh-CN&oi=ao), [Jun Wan](http://www.cbsr.ia.ac.cn/users/jwan/), [Zhen Lei](http://www.cbsr.ia.ac.cn/users/zlei/)

## 📨 Introduction

RAM is an efficient matching framework for OVMLR (Open-Vocabulary Multi-Label Recognition). To address the urgent problems in existing methods, RAM involves (1) LLA to recover regional semantics, and (2) KCOT to find precise region-to-label matching.

## 🔧 Installation

Install the environment through conda and pip is recommended:

```shell
conda create -n ram python=3.10
conda activate ram

# Install the dependencies
pip install -r requirements.txt
```

## 🎯 Running the code
- `model/model.py`: Implementation of RAM model
- `model/ot_solver.py`: Implementation of Sinkhorn Algorithm
- `clip/adapters.py`: Implementation of LLA (Local Adapter)
- `loss/mmc_loss.py`: Implementation of MMC loss (Multi-Matching loss)

Run the following code to start training:
```shell
python train.py --config_file configs/coco.yml
```
Use wandb to log the running:
```shell
python train.py --config_file configs/coco.yml WANDB True
```

## 💬 Discussion
The core contribution is the OT-based matching pipeline, which we found beneficial to the OVMLR task while remaining highly efficient.
The matching framework can be easily extended to dense prediction tasks (e.g., **semantic segmentation**). Welcome to transfer our approach to the segmentation scenarios.

If you find our work useful, please cite our paper:

```
@inproceedings{tan2025recoverandmatch,
  title={Recover and Match: Open-Vocabulary Multi-Label Recognition through Knowledge-Constrained Optimal Transport},
  author={Hao Tan, Zichang Tan, Jun Li, Ajian Liu, Jun Wan, Zhen Lei},
  booktitle={IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year={2025}
}
```

## Acknowledgements

This repo benefits from [MaPLe](https://github.com/muzairkhattak/multimodal-prompt-learning), [CLIP-Surgery](https://github.com/xmed-lab/CLIP_Surgery) and  [POT](https://github.com/PythonOT/POT). Thanks for their wonderful works.

RAM

Owner Repos

File Explorer

# Use via CDN

Was this content helpful?

Recent Feedback

RAM

Owner Repos

File Explorer

# Use via CDN

Was this content helpful?

Recent Feedback

Download README