AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

Joint Visual Semantic Reasoning: Multi-Stage Decoder for Text Recognition

Ayan Kumar Bhunia, Aneeshan Sain, Amandeep Kumar, Shuvozit Ghose, Pinaki Nath Chowdhury, Yi-Zhe Song

ICCV 2021 · Vision-language and multimodal › Human and embodied modeling · approach: Language-grounded scene text and structure · 69 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 Gumbel-softmax differentiable semantic space instead of logits (ablation)Proposed (Stage-2) · IC15word recognition accuracy: 82.8 -> 84.0Improved
add Gumbel-softmax instead of straight-through estimator (ablation)Proposed (Stage-2) · IC15word recognition accuracy: 80.1 -> 84.0Improved
add refinement stages beyond a single stage (s=1 to s=2) (ablation)Proposed (Stage-2) · IC15word recognition accuracy: 80.3 -> 84.0Improved
add a third refinement stage (s=3) (ablation)Proposed (Stage-2) · IC15word recognition accuracy: 84.0 -> 83.6No clear effect
add multi-scale feature maps B_{L-s} per stage (ablation)Proposed (Stage-2) · IC15word recognition accuracy: 81.3 -> 84.0Improved
use FPN lateral connections instead of pyramid pooling (ablation)Proposed (Stage-2) · IC15word recognition accuracy: 81.9 -> 84.0Improved
add visual reasoning block (ablation)Proposed (Stage-2) · IC15word recognition accuracy: 2.9% drop when removedImproved
add semantic reasoning block (ablation)Proposed (Stage-2) · IC15word recognition accuracy: 4.8% drop when removedImproved
add dense connection between glimpse vectors (ablation)Proposed (Stage-2) · IC15word recognition accuracy: 1.6% drop when removedImproved
add residual connection between first and last stage (ablation)Proposed (Stage-2) · IC15word recognition accuracy: 1.3% drop when removedImproved
add multi-stage loss plus auxiliary visual and semantic losses over last-stage loss only (ablation)Proposed (Stage-2) · IC15word recognition accuracy: 73.1 -> 84.0Improved
use Transformer instead of Bi-LSTM reasoning modules (ablation)Proposed (Stage-2) · IC15word recognition accuracy: 81.8 -> 83.1Improved

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