Mo-nan
/

SQKD-model-zoo

English

Knowledge Distillation

Instance Segmentation

Model card Files Files and versions

xet

Community

Mo-nan commited on Jul 17, 2025

Commit

46296b6

verified ·

1 Parent(s): 50a64ef

Update README.md

Browse files

Files changed (1) hide show

README.md +16 -16

README.md CHANGED Viewed

@@ -8,22 +8,22 @@ tags:
 ---
 # Soft Query Knowledge Distillation
 This model provides the main distillation experiment results on dataset ScannetV2 and S3DIS for the 3D instance segmentation task in the model zoo, along with the inference code for the [Mask3D](https://github.com/JonasSchult/Mask3D) model and the [SPFormer3D](https://github.com/sunjiahao1999/SPFormer) method. The table below summarizes the main distillation results from our experiments. You can click on ckpt to download the corresponding Baseline model and the distilled model for each experimental group.
-| ID  | Baseline mAP / mAP50 / mAP25| SQKD mAP | SQKD mAP50 | SQKD mAP25 | download |
-|-----|-----------------------------|----------|------------|------------|----------|
-| 1   | 24.6 / 44.5 / 65.4          | 30.6     | 52.0       | 70.6       | [ckpt1](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/01.zip)|
-| 2   | 37.7 / 63.0 / 79.0          | 41.0     | 66.3       | 81.2       | [ckpt2](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/02.zip)|
-| 3   | 29.7 / 48.8 / 66.5          | 32.2     | 53.8       | 71.1       | [ckpt3](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/03.zip)|
-| 4   | 43.0 / 68.2 / 81.9          | 46.0     | 71.4       | 84.0       | [ckpt4](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/04.zip)|
-| 5   | 42.2 / 68.9 / 83.1          | 47.2     | 73.7       | 86.4       | [ckpt5](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/05.zip)|
-| 6   | 42.4 / 68.0 / 82.3          | 47.6     | 72.6       | 84.1       | [ckpt6](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/06.zip)|
-| 7   | 36.8 / 54.0 / 66.6          | 41.1     | 58.6       | 71.1       | [ckpt7](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/07.zip)|
-| 8   | 38.3 / 56.3 / 68.4          | 41.7     | 59.6       | 70.8       | [ckpt8](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/08.zip)|
-| 9   | 37.2 / 54.9 / 66.6          | 40.1     | 57.6       | 68.0       | [ckpt9](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/09.zip)|
-| 10  | 25.1 / 40.6 / 53.1          | 30.3     | 45.8       | 57.7       | [ckpt10](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/10.zip)|
-| 11  | 19.9 / 27.1 / 39.4          | 23.2     | 30.8       | 42.0       | [ckpt11](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/11.zip)|
-| 12  | 24.5 / 37.8 / 52.4          | 30.8     | 43.3       | 56.9       | [ckpt12](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/12.zip)|
-| 13  | 22.0 / 30.8 / 45.4          | 27.4     | 36.4       | 47.6       | [ckpt13](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/13.zip)|
-| 14  | 33.3 / 44.3 / 55.3          | 38.9     | 49.9       | 59.4       | [ckpt14](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/14.zip)|
 In the table, experimental groups 1–6, 7–10, and 11–14 use different [teacher models](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/teachers.zip), respectively. We also fixed some bugs in the label category code during data preprocessing and provide the improved versions of the [Mask3D](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/code_mix3d.zip) and [SPFormer](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/code_SPFormer.zip) models.

 ---
 # Soft Query Knowledge Distillation
 This model provides the main distillation experiment results on dataset ScannetV2 and S3DIS for the 3D instance segmentation task in the model zoo, along with the inference code for the [Mask3D](https://github.com/JonasSchult/Mask3D) model and the [SPFormer3D](https://github.com/sunjiahao1999/SPFormer) method. The table below summarizes the main distillation results from our experiments. You can click on ckpt to download the corresponding Baseline model and the distilled model for each experimental group.
+| ID  | Baseline mAP / mAP50 / mAP25|Comp.Ratio|   Flops  | SQKD mAP | SQKD mAP50 | SQKD mAP25 | download |
+|-----|-----------------------------|----------|----------|----------|------------|------------|----------|
+| 1   | 24.6 / 44.5 / 65.4          |  *4.17   |   83%    | 30.6     | 52.0       | 70.6       | [ckpt1](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/01.zip)|
+| 2   | 37.7 / 63.0 / 79.0          |  *4.17   |   83%    | 41.0     | 66.3       | 81.2       | [ckpt2](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/02.zip)|
+| 3   | 29.7 / 48.8 / 66.5          |  *2.32   |   95%    | 32.2     | 53.8       | 71.1       | [ckpt3](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/03.zip)|
+| 4   | 43.0 / 68.2 / 81.9          |  *2.32   |   95%    | 46.0     | 71.4       | 84.0       | [ckpt4](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/04.zip)|
+| 5   | 42.2 / 68.9 / 83.1          |  *2.32   |   76%    | 47.2     | 73.7       | 86.4       | [ckpt5](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/05.zip)|
+| 6   | 42.4 / 68.0 / 82.3          |  *2.32   |   57%    | 47.6     | 72.6       | 84.1       | [ckpt6](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/06.zip)|
+| 7   | 36.8 / 54.0 / 66.6          |  *2.44   |   91%    | 41.1     | 58.6       | 71.1       | [ckpt7](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/07.zip)|
+| 8   | 38.3 / 56.3 / 68.4          |  *2.44   |   91%    | 41.7     | 59.6       | 70.8       | [ckpt8](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/08.zip)|
+| 9   | 37.2 / 54.9 / 66.6          |  *2.44   |   71%    | 40.1     | 57.6       | 68.0       | [ckpt9](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/09.zip)|
+| 10  | 25.1 / 40.6 / 53.1          |  *3.87   |   80%    | 30.3     | 45.8       | 57.7       | [ckpt10](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/10.zip)|
+| 11  | 19.9 / 27.1 / 39.4          |  *4.17   |   83%    | 23.2     | 30.8       | 42.0       | [ckpt11](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/11.zip)|
+| 12  | 24.5 / 37.8 / 52.4          |  *4.17   |   83%    | 30.8     | 43.3       | 56.9       | [ckpt12](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/12.zip)|
+| 13  | 22.0 / 30.8 / 45.4          |  *2.32   |   95%    | 27.4     | 36.4       | 47.6       | [ckpt13](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/13.zip)|
+| 14  | 33.3 / 44.3 / 55.3          |  *2.32   |   95%    | 38.9     | 49.9       | 59.4       | [ckpt14](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/14.zip)|
 In the table, experimental groups 1–6, 7–10, and 11–14 use different [teacher models](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/Compress_Result/teachers.zip), respectively. We also fixed some bugs in the label category code during data preprocessing and provide the improved versions of the [Mask3D](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/code_mix3d.zip) and [SPFormer](https://huggingface.co/Mo-nan/SQKD-model-zoo/resolve/main/code_SPFormer.zip) models.