sam2-hiera-tiny
| Property | Value |
|---|---|
| License | Apache 2.0 |
| Downloads | 29,951 |
| Paper | SAM 2: Segment Anything in Images and Videos |
| Tags | Mask Generation, SAM2 |
What is sam2-hiera-tiny?
sam2-hiera-tiny is a lightweight variant of Facebook's SAM2 (Segment Anything Model 2) foundation model, designed for promptable visual segmentation in both images and videos. This tiny version maintains the core functionality while offering a more efficient implementation for resource-conscious applications.
Implementation Details
The model supports both image and video prediction through dedicated predictors (SAM2ImagePredictor and SAM2VideoPredictor). It operates with CUDA acceleration and bfloat16 precision for optimal performance.
- Supports real-time mask generation for images
- Enables video segmentation with frame propagation
- Implements efficient prompt-based segmentation workflow
Core Capabilities
- Image-based mask generation with point or box prompts
- Video segmentation with temporal consistency
- Real-time prompt addition and mask generation
- Frame-by-frame mask propagation in videos
Frequently Asked Questions
Q: What makes this model unique?
This model represents the tiny variant of SAM2, offering a balance between performance and resource efficiency. It's particularly suitable for applications where computational resources are limited but high-quality segmentation is still required.
Q: What are the recommended use cases?
The model is ideal for interactive segmentation tasks in both images and videos, particularly useful in applications requiring real-time performance. It's well-suited for development environments, prototyping, and scenarios where a lighter model footprint is preferred.
