---
title: Sec-4B
canonical_url: "https://www.modelscope.cn/models/AI-ModelScope/Sec-4B"
md_url: "https://www.modelscope.cn/models/AI-ModelScope/Sec-4B.md"
repository: AI-ModelScope/Sec-4B
last_updated: 2025-10-17
license: apache-2.0
pipeline_tag: image-segmentation
tasks:
  - image-segmentation
base_model:
  - OpenIXCLab/SeC-4B
base_model_relation: finetune
parameters: 15.8B
tensor_type:
  - BF16
  - F16
  - F8_E4M3
  - F32
library_name:
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - en
downloads: 1747
stars: 3
---

# Sec-4B

> Sec-4B - AI-ModelScope 在 ModelScope 开源的模型。SeC-4B Model Files - Multiple Precision Formats

AI-ModelScope/Sec-4B 是 ModelScope 魔搭社区上的 15.8B 参数image-segmentation模型，采用 apache-2.0 许可，基于 OpenIXCLab/SeC-4B 构建。

- **Repository**: AI-ModelScope/Sec-4B
- **License**: apache-2.0
- **Tasks**: image-segmentation
- **Parameters**: 15.8B
- **Base model**: OpenIXCLab/SeC-4B
- **Downloads**: 1747
- **Stars**: 3
- **Last updated**: 2025-10-17

Source: https://www.modelscope.cn/models/AI-ModelScope/Sec-4B

---

# SeC-4B Model Files - Multiple Precision Formats

Single-file model formats for the **SeC (Segment Concept)** video object segmentation model, optimized for use with [ComfyUI SeC Nodes](https://github.com/9nate-drake/Comfyui-SecNodes).

## Model Formats

| Format | Size | Description | GPU Requirements |
|--------|------|-------------|------------------|
| **SeC-4B-fp16.safetensors** | 7.35 GB | Recommended - Best balance of quality and size | All CUDA GPUs |
| **SeC-4B-fp8.safetensors** | 3.97 GB | VRAM-constrained systems (saves 1.5-2GB VRAM) | RTX 30 series or newer |
| **SeC-4B-bf16.safetensors** | 7.35 GB | Alternative to FP16 | All CUDA GPUs |
| **SeC-4B-fp32.safetensors** | 14.14 GB | Full precision | All CUDA GPUs |

## What is SeC?

**SeC (Segment Concept)** uses Large Vision-Language Models for video object segmentation, achieving **+11.8 points** improvement over SAM 2.1 on complex semantic scenarios (SeCVOS benchmark).

Key features:
- Concept-driven tracking with semantic understanding
- Handles occlusions and appearance changes
- Bidirectional tracking support
- State-of-the-art performance on multiple benchmarks

## Usage

These models are designed for use with the [ComfyUI SeC Nodes](https://github.com/9nate-drake/Comfyui-SecNodes) custom nodes.

**Installation:**
1. Download your preferred model format
2. Place in `ComfyUI/models/sams/`
3. Install [ComfyUI SeC Nodes](https://github.com/9nate-drake/Comfyui-SecNodes)
4. The model will be automatically detected and available in the SeC Model Loader

## Original Model

These are converted single-file versions of the original model:
- **Original Repository**: [OpenIXCLab/SeC-4B](https://huggingface.co/OpenIXCLab/SeC-4B)
- **Paper**: [arXiv:2507.15852](https://arxiv.org/abs/2507.15852)
- **Official Implementation**: [github.com/OpenIXCLab/SeC](https://github.com/OpenIXCLab/SeC)

## Credits

**Original Model**: Developed by OpenIXCLab
- Model architecture and weights: Apache 2.0 License
- Paper: Zhang et al. "SeC: Advancing Complex Video Object Segmentation via Progressive Concept Construction"

**Single-File Conversions**: Created for [ComfyUI SeC Nodes](https://github.com/9nate-drake/Comfyui-SecNodes)
- Conversion script and ComfyUI integration: [9nate-drake](https://github.com/9nate-drake)
- FP8 quantization support via [torchao](https://github.com/pytorch/ao)

## License

Apache 2.0 (same as original SeC-4B model)

## Citation

If you use this model in your research, please cite the original SeC paper:

```bibtex
@article{zhang2025sec,
  title     = {SeC: Advancing Complex Video Object Segmentation via Progressive Concept Construction},
  author    = {Zhixiong Zhang and Shuangrui Ding and Xiaoyi Dong and Songxin He and
               Jianfan Lin and Junsong Tang and Yuhang Zang and Yuhang Cao and
               Dahua Lin and Jiaqi Wang},
  journal   = {arXiv preprint arXiv:2507.15852},
  year      = {2025}
}
```