Research map / Architectures and efficiency / Quantization
CAR-SAM: Cross-Attention Reconstruction for Post-Training Quantization of the Segment Anything Model
Houji Wen, Jiangyong Yu, Dawei Yang, Jun Li
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add MatMul-Aware Compensation (MAC) (ablation) | SAM-B PTQ · COCO instance segmentation W4A4 | mAP: 28.0 -> 31.2 | Improved |
| add Joint Cross-Attention Reconstruction (JCAR) (ablation) | SAM-B PTQ · COCO instance segmentation W4A4 | mAP: 28.0 -> 35.4 | Improved |
| add MAC and JCAR (ablation) | SAM-B PTQ · COCO instance segmentation W4A4 | mAP: 28.0 -> 39.3 | Improved |
| add MAC (ablation) | SAM-B PTQ · COCO instance segmentation W6A6 | mAP: 50.7 -> 52.7 | Improved |
| add JCAR (ablation) | SAM-B PTQ · COCO instance segmentation W6A6 | mAP: 50.7 -> 53.0 | Improved |
| add MAC and JCAR (ablation) | SAM-B PTQ · COCO instance segmentation W6A6 | mAP: 50.7 -> 53.3 | Improved |
| use CAR-SAM instead of PTQ4SAM | SAM-B 4-bit PTQ · COCO instance segmentation W4A4 | mAP: 24.7 -> 39.3 | Improved |
| use CAR-SAM instead of PTQ4SAM | SAM-L 4-bit PTQ · COCO instance segmentation W4A4 | mAP: 41.9 -> 48.5 | Improved |
| use CAR-SAM instead of PTQ4SAM | SAM2-L 4-bit PTQ · COCO instance segmentation W4A4 | mAP: 38.8 -> 38.2 | No clear effect |
| use CAR-SAM instead of PTQ4SAM | SAM-B 4-bit PTQ · COCO object detection W4A4 | mAP: 50.8 -> 60.6 | Improved |
| use CAR-SAM instead of PTQ4SAM | SAM2-B+ 4-bit PTQ · DAVIS video object segmentation W4A4 | J&F: 71.89 -> 70.18 | Hurt |
| use CAR-SAM instead of PTQ4SAM | SAM2-L 4-bit PTQ · COCO object detection W4A4 (SAM2) | mAP: 65.6 -> 64.1 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- QDrop — QDrop: Randomly Dropping Quantization for Extremely Low-bit Post-Training Quantization (ICLR 2022) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- PTQ4SAM: Post-Training Quantization for Segment Anything (CVPR 2024 · similarity 0.92)
- BA-SAM: Scalable Bias-Mode Attention Mask for Segment Anything Model (CVPR 2024 · similarity 0.89)
- TinySAM: Pushing the Envelope for Efficient Segment Anything Model (AAAI 2025 · similarity 0.88)
- SAM-PARSER: Fine-Tuning SAM Efficiently by Parameter Space Reconstruction (AAAI 2024 · similarity 0.88)
- MimiQ: Low-Bit Data-Free Quantization of Vision Transformers with Encouraging Inter-Head Attention Similarity (AAAI 2025 · similarity 0.87)
- EfficientSAM: Leveraged Masked Image Pretraining for Efficient Segment Anything (CVPR 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
