AI Paper Atlas

Research map / Benchmarks / ROxf Hard

ROxf Hard leaderboard

13 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on ROxf Hard. Best mAP: 78.2 (LPMT, ICML 2025).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

ROxf Hard · mAP

13 papers, each with its best reported number for its own method; higher is better.

#mAP ↑MethodPaperTopic
178.2LPMT
DOLG
Locality Preserving Markovian Transition for Instance Retrieval
ICML 2025 · Table 2
Fast diffusion sampling
270.3Hypergraph Propagation (Ours)
DELG
Hypergraph Propagation and Community Selection for Objects Retrieval
NeurIPS 2021 · Table 1
CNN features
366.57R101-Ours
ResNet101
Learning Token-Based Representation for Image Retrieval
AAAI 2022 · Table 1
Human and embodied modeling
464.8CAS
R-GeM
Cluster-Aware Similarity Diffusion for Instance Retrieval
ICML 2024 · Table: Evaluation of the Perfo
Graph algorithms and clustering
564.2LOCORE-base (DELG RN50)
RN50-DELG
LOCORE: Image Re-ranking with Long-Context Sequence Modeling
CVPR 2025 · Table 1
Autoregressive and latent generation
664.2RRT (top-400)
DELG R50 v2-clean
Instance-Level Image Retrieval Using Reranking Transformers
ICCV 2021 · Table 3
CNN features
762.2MDA-ASMK* (n=2000, CT=128)
ResNet50
Learning Deep Local Features With Multiple Dynamic Attentions for Large-Scale Image Retrieval
ICCV 2021 · Table 4
Image restoration
861.9CiDeR-FT
ResNet101
On Train-Test Class Overlap and Detection for Image Retrieval
CVPR 2024 · Table 6
CNN features
958.97CSD (EfficientNetB3)
R101-DELG
Contextual Similarity Distillation for Asymmetric Image Retrieval
CVPR 2022 · Table 6
Recognition and long-tail
1049.4DSM + diffusion
ResNet101-MAC retrained, upsampled
Local Features and Visual Words Emerge in Activations
CVPR 2019 · Table: state-of-the-art
Autoregressive and latent generation
3 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.