AI Briefing#ocr
The latest AI and developer news about #ocr, with the original source and a short summary.
The latest AI and developer news about #ocr, with the original source and a short summary.
PaddleOCR 3.5 has added Transformers backend support.
Ovis2.6-80B-A3B has been released as an MoE-based multimodal model.
Interfaze has unveiled a new model architecture and API for OCR, STT, and structured output.
DharmaOCR released specialized 3B and 7B SLMs along with cost and performance benchmarks.
Turbo-OCR added a PP-StructureV3 layout model and multilingual OCR.
Pocket LLM v1.5.0 has been released, supporting voice and image input as well as OCR.
Pocket LLM v1.5.0 has been released, adding voice and image input, OCR, and camera capture to Android offline chat.
In a benchmark of 42 documents and 7,560 calls, cheaper/older models outperformed on OCR performance and cost efficiency.
They trained a fast and accurate 6-language OCR model on 12.2 million synthetic data images.
By specializing 3B and 7B open-source SLMs with SFT+DPO, they achieved higher scores than commercial models.
TurboOCR claims up to 1,200 img/s OCR using C++/CUDA and FP16 TensorRT.
A Windows OCR translator built with Rust/Slint, supporting Gemini and local models.
ElevenLabs offers a tool that converts PDF documents into high-quality audio using OCR and TTS technology.
A low-cost OCR stack along with free credits and research support sources was shared for processing 64M pages.
Samsung achieved 97%+ accuracy and under 200ms latency across 7 languages with on-device subtitle spotting for TVs.
Word Shape Encoding significantly boosts Hebrew and Arabic PDF parsing quality.
Mistral has released Mistral OCR 3, which significantly enhances handwriting and complex table recognition performance.
A guide summarizing recent open-source OCR technology trends leveraging Vision Language Models (VLMs), along with model selection and usage methods.
Introduces methods for converting RedNote's 3B-parameter OCR model, dots.ocr, to Core ML and MLX to run on the Apple Neural Engine.
By applying the Kanana model, it achieves a 98% receipt recognition rate and automatically recommends settlement categories.
They implemented an in-house Agent that finds Confluence documents via natural language, using workflows and CQL.
Upstage raised $72 million in Series B funding, surpassing $100 million in cumulative funding.
Upstage will optimize its Solar LLM for RNGD in partnership with FuriosaAI.
NVIDIA has released Llama Nemotron Nano, an 8B-scale VLM with enhanced document understanding and OCR performance.