README.md

---
license: apache-2.0
pipeline_tag: image-text-to-text
library_name: transformers
base_model:
  - OpenGVLab/InternVL3-8B
base_model_relation: merge
language:
  - multilingual
tags:
  - Sa2VA
  - custom_code
---

# Sa2VA: Marrying SAM2 with LLaVA for Dense Grounded Understanding of Images and Videos

[\[📂 GitHub\]](https://github.com/bytedance/Sa2VA)
[\[📜 Sa2VA paper\]](https://arxiv.org/abs/2501.04001)
[\[🚀 Quick Start\]](#quick-start) 



## Introduction

Sa2VA is an MLLM capable of question answering, visual prompt understanding, and dense object segmentation at both image and video levels. It achieves comparable performance to SOTA MLLMs Qwen2.5-VL and InternVL3 on question-answering benchmarks. Additionally, Sa2VA possesses the visual prompt understanding and dense object segmentation capabilities that SOTA MLLMs Qwen2.5-VL and InternVL3 lack. Sa2VA achieves SOTA performance on both image and video grounding and segmentation benchmarks.

## Sa2VA Family

We built the Sa2VA series based on Qwen2.5/3-VL and InternVL2.5/3. In the following table, we provide some Sa2VA models built on Qwen2.5/3-VL and InternVL3.

| Model Name |                             Base MLLM                              |                                Language Part                                |                        HF Link                        |
|:----------:|:------------------------------------------------------------------:|:---------------------------------------------------------------------------:|:-----------------------------------------------------:|
|  Sa2VA-InternVL3-2B  | [InternVL3-2B](https://huggingface.co/OpenGVLab/InternVL3-2B)  | [Qwen2.5-1.5B](https://huggingface.co/Qwen/Qwen2.5-1.5B)  | [🤗 link](https://huggingface.co/ByteDance/Sa2VA-InternVL3-2B)  |
|  Sa2VA-InternVL3-8B  | [InternVL3-8B](https://huggingface.co/OpenGVLab/InternVL3-8B)  |   [Qwen2.5-7B](https://huggingface.co/Qwen/Qwen2.5-7B)    | [🤗 link](https://huggingface.co/ByteDance/Sa2VA-InternVL3-8B)  |
|  Sa2VA-InternVL3-14B  | [InternVL3-14B](https://huggingface.co/OpenGVLab/InternVL3-14B)  | [Qwen2.5-14B](https://huggingface.co/Qwen/Qwen2.5-14B)  | [🤗 link](https://huggingface.co/ByteDance/Sa2VA-InternVL3-14B)  |
| Sa2VA-Qwen2_5-VL-3B  | [Qwen2.5-VL-3B-Instruct](https://huggingface.co/Qwen/Qwen2.5-VL-3B-Instruct) | [Qwen2.5-3B](https://huggingface.co/Qwen/Qwen2.5-3B) | [🤗 link](https://huggingface.co/ByteDance/Sa2VA-Qwen2_5-VL-3B) |
| Sa2VA-Qwen2_5-VL-7B  | [Qwen2.5-VL-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct) | [Qwen2.5-7B](https://huggingface.co/Qwen/Qwen2.5-7B) | [🤗 link](https://huggingface.co/ByteDance/Sa2VA-Qwen2_5-VL-7B) |
|  Sa2VA-Qwen3-VL-2B	 | [Qwen3-VL-2B-Instruct](https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct) |  [Qwen3-1.7B](https://huggingface.co/Qwen/Qwen3-1.7B)   | [🤗 link](https://huggingface.co/ByteDance/Sa2VA-Qwen3-VL-2B) |
| Sa2VA-Qwen3-VL-4B  | [Qwen3-VL-4B-Instruct](https://huggingface.co/Qwen/Qwen3-VL-4B-Instruct) | [Qwen3-4B](https://huggingface.co/Qwen/Qwen3-4B) | [🤗 link](https://huggingface.co/ByteDance/Sa2VA-Qwen3-VL-4B) |

## Sa2VA Performance
| Model Name |   MME    | MMBench  | RefCOCO | RefCOCO+ | RefCOCOg | MeVIS (val_u) | DAVIS |
|:----------:|:--------:|:----:|:-------:|:--------:|:--------:|:-------------:|:-----:|
|  Sa2VA-InternVL3-2B  | 1631/559 | 79.8 |  81.4   |   75.7   |   80.3   |     53.9      | 74.5  |
|  Sa2VA-InternVL3-8B  | 1743/633 | 83.0 |  83.3   |   78.9   |   81.8   |     56.4      | 76.3  |
|  Sa2VA-InternVL3-14B  | 1746/724 | 84.3 |  83.6   |   79.9   |   83.6   |     59.2      | 76.6  |
| Sa2VA-Qwen2_5-VL-3B | 1533/572 | 78.4 |  79.6   |   74.0   |   77.1   |     51.6      | 73.4  |
| Sa2VA-Qwen2_5-VL-7B | 1552/676 | 84.5 |  82.4   |   77.5   |   81.5   |     56.4      | 79.4  |
| Sa2VA-Qwen3-VL-2B | 1541/520 | 79.0 |  80.2   |   75.1   |   78.5   |     53.9      | 74.8  |
| Sa2VA-Qwen3-VL-4B | 1660/655 | 86.3 |  81.7   |   77.4   |   80.0   |     57.1      | 75.9  |

## Quick Start

We provide an example code to run `Sa2VA` using `transformers`.

```python
import torch
from transformers import AutoProcessor, AutoModel
from PIL import Image
import numpy as np
import os

# load the model and processor
path = "ByteDance/Sa2VA-Qwen3-VL-4B"
model = AutoModel.from_pretrained(
    path,
    torch_dtype=torch.bfloat16,
    low_cpu_mem_usage=True,
    use_flash_attn=True,
    trust_remote_code=True).eval().cuda()
processor = AutoProcessor.from_pretrained(path, trust_remote_code=True, use_fast=False)

# for image chat
image_path = "/PATH/TO/IMAGE"
text_prompts = "<image>Please describe the image."
image = Image.open(image_path).convert('RGB')
input_dict = {
    'image': image,
    'text': text_prompts,
    'past_text': '',
    'mask_prompts': None,
    'processor': processor,
    }
return_dict = model.predict_forward(**input_dict)
answer = return_dict["prediction"] # the text format answer

# for image chat with segmentation output
image_path = "/PATH/TO/IMAGE"
text_prompts = "<image>Could you please give me a brief description of the image? Please respond with interleaved segmentation masks for the corresponding parts of the answer."
image = Image.open(image_path).convert('RGB')
input_dict = {
    'image': image,
    'text': text_prompts,
    'past_text': '',
    'mask_prompts': None,
    'processor': processor,
    }
return_dict = model.predict_forward(**input_dict)
answer = return_dict["prediction"] # the text format answer
masks = return_dict['prediction_masks']  # segmentation masks, list(np.array(1, h, w), ...)
    
# for chat with visual prompt (mask format) input
mask_prompts = np.load('/PATH/TO/pred_masks.npy') # np.array(n_prompts, h, w)
image_path = "/PATH/TO/IMAGE"
text_prompts = "<image>Can you provide me with a detailed description of the region in the picture marked by region1."
image = Image.open(image_path).convert('RGB')
input_dict = {
    'image': image,
    'text': text_prompts,
    'past_text': '',
    'mask_prompts': mask_prompts,
    'processor': processor,
    }
return_dict = model.predict_forward(**input_dict)
answer = return_dict["prediction"] # the text format answer

# for video chat
video_folder = "/PATH/TO/VIDEO_FOLDER"
images_paths = os.listdir(video_folder)
images_paths = [os.path.join(video_folder, image_path) for image_name in images_paths]
if len(images_paths) > 5:  # uniformly sample 5 frames
    step = (len(images_paths) - 1) // (5 - 1)
    images_paths = [images_paths[0]] + images_paths[1:-1][::step][1:] + [images_paths[-1]]
text_prompts = "<image>Please describe the video."
input_dict = {
    'video': images_paths,
    'text': text_prompts,
    'past_text': '',
    'mask_prompts': None,
    'processor': processor,
}
return_dict = model.predict_forward(**input_dict)
answer = return_dict["prediction"] # the text format answer


# for video chat with segmentation mask output
video_folder = "/PATH/TO/VIDEO_FOLDER"
images_paths = os.listdir(video_folder)
images_paths = [os.path.join(video_folder, image_path) for image_name in images_paths]
text_prompts = "<image>Please segment the person."
input_dict = {
    'video': images_paths,
    'text': text_prompts,
    'past_text': '',
    'mask_prompts': None,
    'processor': processor,
}
return_dict = model.predict_forward(**input_dict)
answer = return_dict["prediction"] # the text format answer
masks = return_dict['prediction_masks']  # segmentation masks, list(np.array(n_frames, h, w), ...)
```

## Citation

If you find this project useful in your research, please consider citing:

```BibTeX
@article{sa2va,
  title={Sa2VA: Marrying SAM2 with LLaVA for Dense Grounded Understanding of Images and Videos},
  author={Yuan, Haobo and Li, Xiangtai and Zhang, Tao and Huang, Zilong Huang and Xu, Shilin and Ji, Shunping and Tong, Yunhai and Qi, Lu and Feng, Jiashi and Yang, Ming-Hsuan},
  journal={arXiv preprint},
  year={2025}
}
```