AI Paper Atlas

Research map / Detection and segmentation / CNN features

IRSAM: Advancing Segment Anything Model for Infrared Small Target Detection

Mingjin Zhang, Yuchun Wang, Jie Guo, Yunsong Li, Xinbo Gao, Jing Zhang

ECCV 2024 · Detection and segmentation › CNN features · approach: Context for hard detection conditions · 239 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 WPMD blocks to the SAM encoder (ablation)IRSAM · NUAA-SIRSTIoU (%): 78.73 -> 80.78Improved
add WPMD blocks to the SAM encoder (ablation)IRSAM · NUAA-SIRSTFa (1e-6): 11.18 -> 3.95Improved
add Granularity-Aware Decoder (GAD) (ablation)IRSAM · NUAA-SIRSTIoU (%): 79.00 -> 80.78Improved
add GAD (ablation)IRSAM · NUAA-SIRSTFa (1e-6): 2.13 -> 3.95Hurt
add both WPMD and GAD to Mobile-SAM baseline (ablation)Mobile-SAM · NUAA-SIRSTIoU (%): 75.84 -> 80.78Improved
increase number of WPMD blocks from 1 to 4 (ablation)IRSAM · NUAA-SIRSTIoU (%): 76.17 -> 79.00Improved
feed both shallow and deep encoder features to GAD (ablation)IRSAM · NUAA-SIRSTIoU (%): 78.64 -> 80.78Improved
WPMD instead of decoder fine-tuning or AdaptFormer on SAM ViT-BSAM ViT-B · NUAA-SIRSTIoU (%): 67.90 -> 74.15Improved
scale SAM backbone from ViT-B to ViT-H with finetuned decoderSAM · NUAA-SIRSTIoU (%): 67.45 -> 70.96Improved
IRSAM lightweight ViT-Tiny vs SAM ViT-H with WPMDIRSAM · NUAA-SIRSTIoU (%): 75.96 -> 80.78Improved

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