PK PyTorchKR Papers
Papers with Code · CC BY-SA 4.0

Image Retrieval

620 평가 기록 80 datasets

Image Retrieval is a fundamental and long-standing computer vision task that involves finding images similar to a given query from a large database. It is often considered a form of fine-grained, instance-level classification. The task is integral to image recognition alongside classification and cross-modal retrieval. By leveraging visual similarity and other criteria, image retrieval enables users to efficiently discover relevant images, making it a crucial tool in applications such as search and recommendation.

<span class="description-source">Extending CLIP for Category-to-image Retrieval in E-commerce</span>

<span style="color:grey; opacity: 0.6">( Image credit: DELF )</span>

개요는 각 항목의 일부입니다. 위 탭에서 전체 기록을 페이지별로 탐색할 수 있습니다.

벤치마크

Papers with Code 아카이브의 데이터셋별 기록 예시입니다. 평가 조건과 지표 방향이 검증되지 않아 순위를 의미하지 않습니다.

Dataset기록된 모델 예시MetricPaperCode
CoLLM (finetuned - BLIP-L/16)
R@1: 45.8R@10: 84.7R@50: 95.8
논문 코드
MMRet-MLLM
mAP@10: 43.4
논문 코드
RTD + LinCIR (CLIP G/14)
(Recall@10+Recall@50)/2: 56.74
논문 코드
SuperGlobal
mAP: 80.2
논문 코드
AMES
mAP: 90.7
논문 코드
AMES
mAP: 89.7
논문 코드
AMES
mAP: 94.9
논문 코드
ViT-L-14 (LAION400M)
Recall@1 (HN-Atom + HN-Comp, SC): 39.44Recall@1 (HN-Atom + HN-Comp, UC): 33.81Recall@1 (HN-Atom, UC): 47.86Recall@1 (HN-Comp, UC): 60.78
논문 코드
DQU-CIR
(Recall@10+Recall@50)/2: 71.77Recall@10: 61.97
- -
MagicLens (CoCa L)
(Recall@10+Recall@50)/2: 52.95
논문 코드
X-VLM (base)
R@1: 86.9R@10: 98.7R@5: 97.3
논문 코드
TMCIR
(Recall@5+Recall_subset@1)/2: 83.46Recall@10: 91.06
논문 -

데이터셋

80개
CIRRCIRCOFashion IQROxford (Hard)ROxford (Medium)RParis (Hard)RParis (Medium)CREPE (Compositional REPresentation Evaluation)Fashion IQImageNet-RFlickr30K 1K testCIRR

논문

라이브러리

Papers with Code 리더보드 row에 연결된 구현 저장소입니다.

github.com/mesnico/TERAN 1 논문
github.com/abhidipbhattacharyya/srl_aware_ret 1 논문
github.com/adlnlp/attention_vl 1 논문
github.com/hthoai/image-text-matching 1 논문
github.com/iammrhelo/pytorch-vqa-dan 1 논문
github.com/idejie/SCAN 1 논문
github.com/JunweiLiang/DualAttentionNetwork 1 논문
github.com/kuanghuei/SCAN 1 논문

하위 태스크

CI Composed Image Retrieval (CoIR) 14 데이터셋 · 206 rows CI Content-Based Image Retrieval 3 데이터셋 · 18 rows SI Sketch-Based Image Retrieval 3 데이터셋 · 17 rows CI Chat-based Image Retrieval 1 데이터셋 · 3 rows II Image Instance Retrieval 1 데이터셋 · 2 rows MI Medical Image Retrieval 1 데이터셋 · 1 rows V Video-to-Shop 1 데이터셋 · 1 rows FI Face Image Retrieval 0 데이터셋 · 0 rows MI Multi-Label Image Retrieval 0 데이터셋 · 0 rows SS Semi-Supervised Sketch Based Image Retrieval 0 데이터셋 · 0 rows