Mindee: Surya

docTR (Document Text Recognition) is an open-source OCR toolkit developed by Mindee, with its initial public release in March 2021 under the Apache 2.0 license. It provides end-to-end document text recognition through a two-stage pipeline consisting of text detection and text recognition, both implemented as deep learning models. docTR supports multiple detection architectures including DBNet and LinkNet, and recognition architectures including CRNN and SAR, with both TensorFlow and PyTorch backends available.docTR is designed for reading text in document images including scanned PDFs, photographs of printed documents, and forms. It handles multilingual text recognition across standard Latin-script languages and is deployable through Roboflow Inference. It is suited for document digitization pipelines, automated form processing, and applications requiring accurate structured text extraction from document images.

TrOCR (Transformer-based Optical Character Recognition) is an end-to-end OCR model released in September 2021 by Microsoft Research. It departs from the traditional two-stage OCR pipeline — which typically combines a CNN-based feature extractor with an RNN-based sequence decoder — by using a pure Transformer architecture composed of a pretrained image Transformer encoder and a pretrained text Transformer decoder, an approach that later became standardized as the VisionEncoderDecoder pattern in Hugging Face Transformers.TrOCR takes a cropped text line image as input and produces a sequence of output tokens, supporting printed, handwritten, and scene text recognition. The model is designed for use downstream of a separate text detection stage — TrOCR recognizes text in pre-cropped regions rather than detecting text locations in a full page. Microsoft released three size variants: TrOCR-small (62M parameters, DeiT-small encoder + MiniLM decoder), TrOCR-base (334M parameters, BEiT-base encoder + RoBERTa-large decoder), and TrOCR-large (558M parameters, BEiT-large encoder + RoBERTa-large decoder). Pretrained and fine-tuned checkpoints are available for printed text (on SROIE), handwritten text (on IAM), and scene text (on the standard scene text benchmarks) under the MIT license, distributed through the Microsoft unilm repository and Hugging Face. At release, TrOCR achieved state-of-the-art results across all three benchmark categories, and the model continues to be used as a baseline for handwritten text recognition.

Florence-2, introduced by Microsoft Research at CVPR 2024, is an open-source vision-language foundation model designed to unify diverse computer vision tasks within a single sequence-to-sequence framework. Unlike traditional models that specialize in specific tasks, Florence-2 accepts both images and text prompts and outputs text for tasks such as captioning, object detection, segmentation, OCR, and region-based grounding. It comes in two sizes—Florence-2-base (~230M parameters) and Florence-2-large (~770M parameters)—and is trained on FLD-5B, a large dataset of ~126M images with ~5.4B annotations.The model demonstrates strong zero-shot and fine-tuned performance, often rivaling larger vision-language systems while remaining lightweight and efficient. Released under the MIT license, all weights are publicly available, making it accessible for fine-tuning and deployment in applications like VQA, content tagging, accessibility, and research. Florence-2’s compact design, versatility, and openness position it as a practical alternative to larger proprietary multimodal models.

PaliGemma 2 is a vision-language model released in December 2024 by Google DeepMind. It pairs the SigLIP-So400m vision encoder with the Gemma 2 language model family, extending the original PaliGemma architecture with stronger language capabilities and a wider set of transfer benchmarks. The model is designed primarily as a fine-tuning base rather than a chat-optimized assistant. Google releases pretrained "PT" checkpoints intended for task-specific adaptation rather than direct out-of-the-box use.PaliGemma 2 accepts an image paired with a text prompt and generates natural language output, supporting image captioning, visual question answering, optical character recognition, document understanding, object detection and segmentation (with appropriate fine-tuning), and a range of specialized vision-language tasks. The model is released at three parameter sizes (3B, 10B, and 28B), built on the Gemma 2 2B, 9B, and 27B language backbones. Each size is available at three input resolutions: 224, 448, and 896 pixels. Alongside the base PT checkpoints, Google released PaliGemma 2 Mix variants that have been tuned on a mixture of downstream tasks to provide stronger out-of-the-box performance for common applications such as OCR and document parsing. PaliGemma 2 is distributed under the Gemma license, a custom license from Google that permits commercial use subject to the terms of the Gemma Prohibited Use Policy.

GLM-OCR is a multimodal OCR model for complex document understanding, built on the GLM-V encoder-decoder architecture by Zhipu AI. The model combines a 0.4B-parameter CogViT visual encoder pre-trained on large-scale image-text data, a lightweight cross-modal connector with efficient token downsampling, and a 0.5B-parameter GLM language decoder, totaling 0.9B parameters. To address the inefficiency of standard autoregressive decoding in deterministic OCR tasks, GLM-OCR introduces a Multi-Token Prediction (MTP) mechanism that predicts multiple tokens per step, significantly improving decoding throughput while keeping memory overhead low through shared parameters. Training proceeds through four stages: visual encoder pretraining with MIM, CLIP, and distillation objectives; vision-language pretraining on document parsing, grounding, and VQA data; supervised fine-tuning on curated OCR datasets covering text, formula, table, and key information extraction; and full-task reinforcement learning to improve accuracy and structural consistency.At the system level, GLM-OCR adopts a two-stage pipeline in which PP-DocLayout-V3 first performs layout analysis, followed by parallel region-level recognition. This design enables robust handling of diverse document layouts including tables, formulas, and multi-column text. The model supports document parsing and targeted recognition tasks, producing structured outputs in Markdown, JSON, and LaTeX formats across more than 100 languages. On the OmniDocBench V1.5 benchmark, GLM-OCR scores 94.62, and achieves 94.0 on OCRBench and 96.5 on UniMERNet for formula recognition.

Qwen2.5-VL-7B-Instruct is a 7-billion parameter vision-language model from Alibaba’s QwenLM team, released on January 26, 2025 under the Apache 2.0 license. It is the instruction-tuned variant of the 7B scale in the Qwen2.5-VL family, designed to process multimodal inputs such as text, images, charts, documents, and video. The model enables structured outputs—including JSON for structured content and bounding boxes for visual localization. Weights are publicly available on Hugging Face and GitHub, making it suitable for both research and applied multimodal use.