Research map / Detection and segmentation / CNN features
DeepSolo: Let Transformer Decoder With Explicit Points Solo for Text Spotting
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 TextOCR to pretraining data (ablation) | DeepSolo (ResNet-50) · Total-Text test | E2E None F1: 79.65 -> 82.54 | Helped |
| Add MLT17, IC13, and IC15 real training data (ablation) | DeepSolo (ResNet-50) · Total-Text test | E2E None F1: +0.82% | Helped |
| Remove the text matching criterion (ablation) | DeepSolo · Total-Text test | E2E None F1: 78.83 -> 77.85 | Hurt |
| Share point embeddings across instances (ablation) | DeepSolo · Total-Text test | E2E None F1: 78.83 -> 77.58 | Hurt |
| Increase maximum recognition length from 25 to 50 (ablation) | DeepSolo (ResNet-50, Synth150K) · CTW1500 test | E2E None: 60.1 -> 63.2 | Helped |
| Replace ResNet-50 with ViTAEv2-S (ablation) | DeepSolo · Total-Text test | E2E None F1: 79.65 -> 81.79 | Helped |
| Set text loss weight to 0.25 instead of 0.5 (ablation) | DeepSolo · Total-Text test | E2E None F1: 78.83 -> 76.68 | Hurt |
| Set text loss weight to 0.75 instead of 0.5 (ablation) | DeepSolo · Total-Text test | E2E None F1: 78.83 -> 78.82 | No clear effect |
| Use line annotations instead of full polygon annotations (ablation) | DeepSolo · Total-Text test | E2E None: 81.6% vs. 82.5% | No clear effect |
Closest papers by meaning
- OTE: Exploring Accurate Scene Text Recognition Using One Token (CVPR 2024 · similarity 0.88)
- Arbitrary Reading Order Scene Text Spotter with Local Semantics Guidance (AAAI 2025 · similarity 0.87)
- SemiETS: Integrating Spatial and Content Consistencies for Semi-Supervised End-to-end Text Spotting (CVPR 2025 · similarity 0.87)
- GoMatching: A Simple Baseline for Video Text Spotting via Long and Short Term Matching (NeurIPS 2024 · similarity 0.87)
- Bridging the Gap Between End-to-End and Two-Step Text Spotting (CVPR 2024 · similarity 0.86)
- InstructOCR: Instruction Boosting Scene Text Spotting (AAAI 2025 · similarity 0.86)
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.