AI Paper Atlas

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

Product1M: Towards Weakly Supervised Instance-Level Product Retrieval via Cross-Modal Pretraining

Xunlin Zhan, Yangxin Wu, Xiao Dong, Yunchao Wei, Minlong Lu, Yichi Zhang, Hang Xu, Xiaodan Liang

ICCV 2021 · Vision-language and multimodal › Cross-modal learning · approach: Probabilistic embeddings for cross-modal retrieval · 57 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 image-text matching (ITM) pretext task (ablation)CAPTURE masked pretraining · Product1M pretext tasks (N=100)mAP: 72.1 -> 70.2Hurt
add cross-modal contrastive loss (ablation)CAPTURE masked pretraining · Product1M pretext tasks (N=100)mAP: 72.1 -> 73.3Improved
concatenate Text/Visual Transformer features with Co-Transformer features for retrieval (ablation)CAPTURE with contrastive loss · Product1M pretext tasks (N=100)mAP: 73.3 -> 74.6Improved
concatenate features without contrastive loss (ablation)CAPTURE masked pretraining · Product1M pretext tasks (N=100)mAP: 72.1 -> 71.9Hurt
add Cross-Transformer layers (ablation)CAPTURE · Product1M layer configuration (N=100)mAP: 73.8 -> 74.6Improved
add Co-Transformer layers (ablation)CAPTURE · Product1M layer configuration (N=100)mAP: 73.2 -> 74.6Improved
add Text/Visual Transformer layers (ablation)CAPTURE · Product1M layer configuration (N=100)mAP: 69.3 -> 74.6Improved
double depth from 12 to 24 layers (CAPTURE-L) (ablation)CAPTURE · Product1M layer configuration (N=100)mAP: 74.6 -> 74.7No clear effect
increase depth from 6 to 12 layers (ablation)CAPTURE · Product1M layer configuration (N=100)mAP: 67.7 -> 74.6Improved
use augmentation-trained RPN instead of Faster R-CNN pretrained on Visual Genome (ablation)CAPTURE · Product1M detection impactmAP@10: 70.36 -> 74.63Improved
use detected instances instead of one whole-image box (ablation)CAPTURE · Product1M detection impactmAP@10: 60.03 -> 74.63Improved
use ground truth boxes instead of RPN boxes (ablation)CAPTURE · Product1M annotated subsetmAP@10: 73.36 -> 77.79Improved

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