Research map / Vision-language and multimodal / Human and embodied modeling
Platypus: A Generalized Specialist Model for Reading Text in Various Forms
Peng Wang, Zhaohai Li, Jun Tang, Humen Zhong, Fei Huang, Zhibo Yang, Cong Yao
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add category and writing type prompts for handwriting (ablation) | Platypus · IAM | accuracy: 88.6 -> 96.4 | Improved |
| add category prompt for formula recognition (ablation) | Platypus · LaTeX-OCR | WER: 8.5 -> 7.8 | Improved |
| add prompts for scene text recognition (ablation) | Platypus · IIIT5k | accuracy: 98.8 -> 99.1 | No clear effect |
| add full-image pre-training before joint training instead of none (ablation) | Platypus · IC15 spotting | H-mean: 0.1 -> 86.7 | Improved |
| pre-train on full images instead of cropped images (ablation) | Platypus · IC15 spotting | H-mean: 61.2 -> 86.7 | Improved |
| increase training image size (768/512 to 1024/768) (ablation) | Platypus · IC15 spotting | H-mean: 82.4 -> 88.2 | Improved |
| use point or box prompts instead of full-image RAT (ablation) | Platypus · TotalText spotting (word-level) | H-mean: 78.9 -> 93.0 | Improved |
| unified Platypus versus GPT-4V on handwriting | Platypus · IAM | word accuracy: 50.7 -> 96.4 | Improved |
| unified model versus specialist PARSeq on scene text recognition | Platypus · STR average | word accuracy: 96.4 -> 96.4 | No clear effect |
Closest papers by meaning
- On the Generalization of Handwritten Text Recognition Models (CVPR 2025 · similarity 0.85)
- OCRBench v2: An Improved Benchmark for Evaluating Large Multimodal Models on Visual Text Localization and Reasoning (NeurIPS 2025 · similarity 0.84)
- OTE: Exploring Accurate Scene Text Recognition Using One Token (CVPR 2024 · similarity 0.84)
- Glimpse: Enabling White-Box Methods to Use Proprietary Models for Zero-Shot LLM-Generated Text Detection (ICLR 2025 · similarity 0.83)
- Out of Length Text Recognition with Sub-String Matching (AAAI 2025 · similarity 0.83)
- PP-OCRv5: A Specialized 5M-Parameter Model Rivaling Billion-Parameter Vision-Language Models on OCR Tasks (CVPR 2026 · similarity 0.83)
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.
