Update README.md

README.md

@@ -29,7 +29,8 @@ Segment Anything Model (SAM) has shown impressive zero-shot transfer performance
 
 ## Installation
 ```bash
-pip install -e .
+git clone https://github.com/THU-MIG/RepViT
+cd sam && pip install -e .
 # download pretrained checkpoint
 mkdir weights && cd weights
 wget https://github.com/THU-MIG/RepViT/releases/download/v1.0/repvit_sam.pt
@@ -42,11 +43,11 @@ python app/app.py
 ```
 
 ## CoreML export
-Please refer to [coreml_example.ipynb](./notebooks/coreml_example.ipynb)
+Please refer to [coreml_example.ipynb](https://github.com/THU-MIG/RepViT/blob/main/sam/notebooks/coreml_example.ipynb)
 
 
 ## Latency comparisons
-Comparison between RepViT-SAM and others in terms of latency. The latency (ms) is measured with the standard resolution of 1024 $\times$ 1024 on iPhone 12 and Macbook M1 Pro by Core ML Tools. OOM means out of memory.
+Comparison between RepViT-SAM and others in terms of latency. The latency (ms) is measured with the standard resolution of 1024 x 1024 on iPhone 12 and Macbook M1 Pro by Core ML Tools. OOM means out of memory.
 
 <table class="tg">
 <thead>
@@ -74,195 +75,6 @@ Comparison between RepViT-SAM and others in terms of latency. The latency (ms) i
 </tbody>
 </table>
 
-
-## Zero-shot edge detection
-
-Comparison results on BSDS500.
-
-<table class="tg">
-<thead>
-  <tr>
-    <th class="tg-c3ow" rowspan="2">Model</th>
-    <th class="tg-c3ow" colspan="3">zero-shot edge detection</th>
-  </tr>
-  <tr>
-    <th class="tg-c3ow">ODS</th>
-    <th class="tg-c3ow">OIS</th>
-    <th class="tg-c3ow">AP</th>
-  </tr>
-</thead>
-<tbody>
-  <tr>
-    <td class="tg-c3ow">ViT-H-SAM</td>
-    <td class="tg-c3ow"><b>.768</b></td>
-    <td class="tg-c3ow"><b>.786</b></td>
-    <td class="tg-c3ow"><b>.794</b></td>
-  </tr>
-  <tr>
-    <td class="tg-c3ow">ViT-B-SAM</td>
-    <td class="tg-c3ow">.743</td>
-    <td class="tg-c3ow">.764</td>
-    <td class="tg-c3ow">.726</td>
-  </tr>
-  <tr>
-    <td class="tg-c3ow">MobileSAM</td>
-    <td class="tg-c3ow">.756</td>
-    <td class="tg-c3ow">.768</td>
-    <td class="tg-c3ow">.746</td>
-  </tr>
-  <tr>
-    <td class="tg-c3ow">RepViT-SAM</td>
-    <td class="tg-c3ow"><ins>.764</ins></td>
-    <td class="tg-c3ow"><ins>.786</ins></td>
-    <td class="tg-c3ow"><ins>.773</ins></td>
-  </tr>
-</tbody>
-</table>
-
-
-## Zero-shot instance segmentation and SegInW
-Comparison results on COCO and SegInW.
-
-<table class="tg">
-<thead>
-  <tr>
-    <th class="tg-c3ow" rowspan="2">Model</th>
-    <th class="tg-c3ow" colspan="4">zero-shot instance segmentation</th>
-    <th class="tg-c3ow">SegInW</th>
-  </tr>
-  <tr>
-    <th class="tg-c3ow">AP</th>
-    <th class="tg-c3ow">$AP^{S}$</th>
-    <th class="tg-c3ow">$AP^{M}$</th>
-    <th class="tg-c3ow">$AP^{L}$</th>
-    <th class="tg-c3ow">Mean AP</th>
-  </tr>
-</thead>
-<tbody>
-  <tr>
-    <td class="tg-c3ow">ViT-H-SAM</td>
-    <td class="tg-c3ow"><b>46.8</b></td>
-    <td class="tg-c3ow"><b>31.8</b></td>
-    <td class="tg-c3ow"><b>51.0</b></td>
-    <td class="tg-c3ow"><b>63.6</b></td>
-    <td class="tg-c3ow"><b>48.7</b></td>
-  </tr>
-  <tr>
-    <td class="tg-c3ow">ViT-B-SAM</td>
-    <td class="tg-c3ow">42.5</td>
-    <td class="tg-c3ow"><ins>29.8</ins></td>
-    <td class="tg-c3ow">47.0</td>
-    <td class="tg-c3ow">56.8</td>
-    <td class="tg-c3ow">44.8</td>
-  </tr>
-  <tr>
-    <td class="tg-c3ow">MobileSAM</td>
-    <td class="tg-c3ow">42.7</td>
-    <td class="tg-c3ow">27.0</td>
-    <td class="tg-c3ow">46.5</td>
-    <td class="tg-c3ow">61.1</td>
-    <td class="tg-c3ow">43.9</td>
-  </tr>
-  <tr>
-    <td class="tg-c3ow">RepViT-SAM</td>
-    <td class="tg-c3ow"><ins>44.4</ins></td>
-    <td class="tg-c3ow">29.1</td>
-    <td class="tg-c3ow"><ins>48.6</ins></td>
-    <td class="tg-c3ow"><ins>61.4</ins></td>
-    <td class="tg-c3ow"><ins>46.1</ins></td>
-  </tr>
-</tbody>
-</table>
-
-## Zero-shot video object/instance segmentation
-Comparison results on DAVIS 2017 and UVO.
-
-<table class="tg">
-<thead>
-  <tr>
-    <th class="tg-c3ow" rowspan="2">Model</th>
-    <th class="tg-c3ow" colspan="3">z.s. VOS</th>
-    <th class="tg-c3ow">z.s. VIS</th>
-  </tr>
-  <tr>
-    <th class="tg-c3ow">$\mathcal{J\&amp;F}$</th>
-    <th class="tg-c3ow">$\mathcal{J}$</th>
-    <th class="tg-c3ow">$\mathcal{F}$</th>
-    <th class="tg-c3ow">AR100</th>
-  </tr>
-</thead>
-<tbody>
-  <tr>
-    <td class="tg-c3ow">ViT-H-SAM</td>
-    <td class="tg-c3ow"><b>77.4</b></td>
-    <td class="tg-c3ow"><b>74.6</b></td>
-    <td class="tg-c3ow"><b>80.2</b></td>
-    <td class="tg-c3ow"><b>28.8</b></td>
-  </tr>
-  <tr>
-    <td class="tg-c3ow">ViT-B-SAM</td>
-    <td class="tg-c3ow">71.3</td>
-    <td class="tg-c3ow">68.5</td>
-    <td class="tg-c3ow">74.1</td>
-    <td class="tg-c3ow">19.1</td>
-  </tr>
-  <tr>
-    <td class="tg-c3ow">MobileSAM</td>
-    <td class="tg-c3ow">71.1</td>
-    <td class="tg-c3ow">68.6</td>
-    <td class="tg-c3ow">73.6</td>
-    <td class="tg-c3ow">22.7</td>
-  </tr>
-  <tr>
-    <td class="tg-c3ow">RepViT-SAM</td>
-    <td class="tg-c3ow"><ins>73.5</ins></td>
-    <td class="tg-c3ow"><ins>71.0</ins></td>
-    <td class="tg-c3ow"><ins>76.1</ins></td>
-    <td class="tg-c3ow"><ins>25.3</ins></td>
-  </tr>
-</tbody>
-</table>
-
-## Zero-shot salient object segmentation
-Comparison results on DUTS.
-## Zero-shot anomaly detection
-Comparison results on MVTec.
-<table class="tg">
-<thead>
-  <tr>
-    <th class="tg-c3ow" rowspan="2">Model</th>
-    <th class="tg-c3ow">z.s. s.o.s.</th>
-    <th class="tg-c3ow">z.s. a.d.</th>
-  </tr>
-  <tr>
-    <th class="tg-c3ow">$\mathcal{M}$ $\downarrow$</th>
-    <th class="tg-c3ow">$\mathcal{F}_{p}$</th>
-  </tr>
-</thead>
-<tbody>
-  <tr>
-    <td class="tg-c3ow">ViT-H-SAM</td>
-    <td class="tg-c3ow"><b>0.046</b></td>
-    <td class="tg-c3ow"><ins>37.65</ins></td>
-  </tr>
-  <tr>
-    <td class="tg-c3ow">ViT-B-SAM</td>
-    <td class="tg-c3ow">0.121</td>
-    <td class="tg-c3ow">36.62</td>
-  </tr>
-  <tr>
-    <td class="tg-c3ow">MobileSAM</td>
-    <td class="tg-c3ow">0.147</td>
-    <td class="tg-c3ow">36.44</td>
-  </tr>
-  <tr>
-    <td class="tg-c3ow">RepViT-SAM</td>
-    <td class="tg-c3ow"><ins>0.066</ins></td>
-    <td class="tg-c3ow"><b>37.96</b></td>
-  </tr>
-</tbody>
-</table>
-
 ## Acknowledgement
 
 The code base is partly built with [SAM](https://github.com/facebookresearch/segment-anything) and [MobileSAM](https://github.com/ChaoningZhang/MobileSAM).