Zigeng
/

SlimSAM-uniform-77

@@ -5,15 +5,12 @@ tags:
 ---
 # SlimSAM: 0.1% Data Makes Segment Anything Slim
-<div align="center">
-<img src="images/paper/intro.PNG" width="66%">
-<img src="images/paper/everything.PNG" width="100%">
-</div>
 > **0.1% Data Makes Segment Anything Slim**
 > [Zigeng Chen](https://github.com/czg1225), [Gongfan Fang](https://fangggf.github.io/), [Xinyin Ma](https://horseee.github.io/), [Xinchao Wang](https://sites.google.com/site/sitexinchaowang/)
 > [Learning and Vision Lab](http://lv-nus.org/), National University of Singapore
 > Paper: [[Arxiv]](https://arxiv.org/abs/2312.05284)
 ## Introduction
@@ -23,38 +20,9 @@ tags:
 **SlimSAM** is a novel SAM compression method, which efficiently reuses pre-trained SAMs without the necessity for extensive retraining. This is achieved by the efficient reuse of pre-trained SAMs through a unified pruning-distillation framework. To enhance knowledge inheritance from the original SAM, we employ an innovative alternate slimming strategy that partitions the compression process into a progressive procedure. Diverging from prior pruning techniques, we meticulously prune and distill decoupled model structures in an alternating fashion. Furthermore, a novel label-free pruning criterion is also proposed to align the pruning objective with the optimization target, thereby boosting the post-distillation after pruning.
-![Frame](images/paper/frame.PNG?raw=true)
 SlimSAM achieves approaching performance while reducing the parameter counts to **0.9\% (5.7M)**, MACs to **0.8\% (21G)**, and requiring mere **0.1\% (10k)** of the training data when compared to the original SAM-H. Extensive experiments demonstrate that our method realize significant superior performance while utilizing over **10 times** less training data when compared to other SAM compression methods.
-## Visualization Results
-Qualitative comparison of results obtained using point prompts, box prompts, and segment everything prompts are shown in the following section.
-### Segment Everything Prompts
-<div align="center">
-  <img src="images/paper/everything2.PNG" width="100%">
-</div>
-### Box Prompts and Point Prompts
-<div align="center">
-  <img src="images/paper/prompt.PNG" width="100%">
-</div>
-## Quantitative Results
-We conducted a comprehensive comparison encompassing performance, efficiency, and training costs with other SAM compression methods and structural pruning methods.
-### Comparing with other SAM compression methods.
-<div align="center">
-  <img src="images/paper/compare_tab1.PNG" width="100%">
-</div>
-### Comparing with other structural pruning methods.
-<div align="center">
-  <img src="images/paper/compare_tab2.PNG" width="50%">
-</div>

 ---
 # SlimSAM: 0.1% Data Makes Segment Anything Slim
 > **0.1% Data Makes Segment Anything Slim**
 > [Zigeng Chen](https://github.com/czg1225), [Gongfan Fang](https://fangggf.github.io/), [Xinyin Ma](https://horseee.github.io/), [Xinchao Wang](https://sites.google.com/site/sitexinchaowang/)
 > [Learning and Vision Lab](http://lv-nus.org/), National University of Singapore
 > Paper: [[Arxiv]](https://arxiv.org/abs/2312.05284)
+> Code: [[GitHub]](https://github.com/czg1225/SlimSAM)
 ## Introduction
 **SlimSAM** is a novel SAM compression method, which efficiently reuses pre-trained SAMs without the necessity for extensive retraining. This is achieved by the efficient reuse of pre-trained SAMs through a unified pruning-distillation framework. To enhance knowledge inheritance from the original SAM, we employ an innovative alternate slimming strategy that partitions the compression process into a progressive procedure. Diverging from prior pruning techniques, we meticulously prune and distill decoupled model structures in an alternating fashion. Furthermore, a novel label-free pruning criterion is also proposed to align the pruning objective with the optimization target, thereby boosting the post-distillation after pruning.
 SlimSAM achieves approaching performance while reducing the parameter counts to **0.9\% (5.7M)**, MACs to **0.8\% (21G)**, and requiring mere **0.1\% (10k)** of the training data when compared to the original SAM-H. Extensive experiments demonstrate that our method realize significant superior performance while utilizing over **10 times** less training data when compared to other SAM compression methods.