AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

MolSight: Optical Chemical Structure Recognition with SMILES Pretraining, Multi-Granularity Learning and Reinforcement Learning

Wenrui Zhang, Xinggang Wang, Bin Feng, Wenyu Liu

AAAI 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Dense features and foundation priors · 4 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 chemical bond head to SMILES decoder (ablation)MolSight · USPTOgraph accuracy: 93.7 -> 94.2Improved
jointly train atom localization with SMILES and bond tasks (ablation)MolSight · USPTOstereo accuracy: 10.6 vs 80.9 (SMILES+edge)Hurt
train coordinate head separately from other tasks (ablation)MolSight · USPTOstereo accuracy: 10.6 -> 29.4Improved
add two extra decoder layers for coordinate branch (L1 loss) (ablation)MolSight coordinate head · USPTOstereo accuracy: 29.4 -> 82.4Improved
use MLE (Laplace) loss instead of L1 loss for coordinates (ablation)MolSight coordinate head · USPTOstereo accuracy: 82.4 -> 83.7Improved
RL post-training with Tanimoto reward only (ablation)MolSight SFT · Stereo-2kstereo accuracy: 80.1 -> 81.8Improved
RL post-training with weighted reward (ablation)MolSight SFT · Stereo-2kstereo accuracy: 80.1 -> 87.1Improved
RL post-training with weighted reward (ablation)MolSight SFT · CLEF-2012 (out of domain)stereo accuracy: 71.0 -> 80.6Improved
RL post-training with weighted reward (ablation)MolSight SFT · CLEF-2012 (out of domain)Tanimoto: 90.0 -> 90.0No clear effect
use stereo reward alone vs weighted reward in RL (ablation)MolSight RL · Stereo-2kstereo accuracy: 87.1 -> 86.5Hurt
MolSight pretraining of image encoder instead of ImageNet pretraining (ablation)EfficientViT · MoleculeNet BBBPROC-AUC: 63.6 -> 68.0Improved
MolSight vs MolScribe fine-tuned 30 epochsMolSight · CLEFexact match accuracy: 85.5 -> 84.9No clear effect

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