AI Paper Atlas

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

CVPR 2026 · Architectures and efficiency › Quantization · approach: Post-training quantization of ViTs · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
add MatMul-Aware Compensation (MAC) (ablation)SAM-B PTQ · COCO instance segmentation W4A4mAP: 28.0 -> 31.2Improved
add Joint Cross-Attention Reconstruction (JCAR) (ablation)SAM-B PTQ · COCO instance segmentation W4A4mAP: 28.0 -> 35.4Improved
add MAC and JCAR (ablation)SAM-B PTQ · COCO instance segmentation W4A4mAP: 28.0 -> 39.3Improved
add MAC (ablation)SAM-B PTQ · COCO instance segmentation W6A6mAP: 50.7 -> 52.7Improved
add JCAR (ablation)SAM-B PTQ · COCO instance segmentation W6A6mAP: 50.7 -> 53.0Improved
add MAC and JCAR (ablation)SAM-B PTQ · COCO instance segmentation W6A6mAP: 50.7 -> 53.3Improved
use CAR-SAM instead of PTQ4SAMSAM-B 4-bit PTQ · COCO instance segmentation W4A4mAP: 24.7 -> 39.3Improved
use CAR-SAM instead of PTQ4SAMSAM-L 4-bit PTQ · COCO instance segmentation W4A4mAP: 41.9 -> 48.5Improved
use CAR-SAM instead of PTQ4SAMSAM2-L 4-bit PTQ · COCO instance segmentation W4A4mAP: 38.8 -> 38.2No clear effect
use CAR-SAM instead of PTQ4SAMSAM-B 4-bit PTQ · COCO object detection W4A4mAP: 50.8 -> 60.6Improved
use CAR-SAM instead of PTQ4SAMSAM2-B+ 4-bit PTQ · DAVIS video object segmentation W4A4J&F: 71.89 -> 70.18Hurt
use CAR-SAM instead of PTQ4SAMSAM2-L 4-bit PTQ · COCO object detection W4A4 (SAM2)mAP: 65.6 -> 64.1Hurt

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.

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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy