AI Paper Atlas

Research map / Vision-language and multimodal / Cross-modal learning

Intra-class Distribution-guided Generative Hashing with Neighbor Refinement for Cross-modal Retrieval

Hao Sun, Yadong Huo, Qibing Qin, Wenfeng Zhang, Lei Huang

CVPR 2026 · Vision-language and multimodal › Cross-modal learning · approach: Cross-modal hashing

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 DAG sample generation (ablation)IDGH baseline · XMediaNet I->T (16 bits)mAP: 0.1662 -> 0.3928Improved
add NDR neighbor refinement on top of DAG (ablation)IDGH · XMediaNet I->T (16 bits)mAP: 0.3928 -> 0.4046Improved
add NDR neighbor refinement on top of DAG (ablation)IDGH · NUS-WIDE I->T (128 bits)mAP: 0.7470 -> 0.7565Improved
add NDR neighbor refinement on top of DAG (ablation)IDGH · MIRFLICKR-25K I->T (16 bits)mAP: 0.8434 -> 0.8392Hurt
add IDGH to contrastive lossContrastive loss · MIRFLICKR-25K I->T (16 bits)mAP: 0.7414 -> 0.7581Improved
add IDGH to triplet lossTriplet loss · MIRFLICKR-25K I->T (16 bits)mAP: 0.8233 -> 0.8392Improved
add IDGH to multi-similarity lossMulti-similarity loss · MIRFLICKR-25K I->T (16 bits)mAP: 0.7791 -> 0.7991Improved
add IDGH to proxy-NCA lossProxy-NCA loss · MIRFLICKR-25K I->T (16 bits)mAP: 0.7644 -> 0.8022Improved
use IDGH generation instead of MixupIDGH · MIRFLICKR-25K T->I (128 bits)mAP: 0.8340 -> 0.8411Improved
use IDGH generation instead of HashGANIDGH · MIRFLICKR-25K I->T (16 bits)mAP: 0.8309 -> 0.8392Improved
increase number of synthetic samples M beyond 10 (ablation)IDGH · IAPR TC-12 (128 bits)mAPHurt
increase number of neighbor classes K (ablation)IDGH · IAPR TC-12 (128 bits)mAPImproved

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