trocr-large-printed

Name: trocr-large-printed
Author: microsoft

by microsoft

Transformer-based OCR for printed text recognition

65Kdl/month

181likes

608Mparams

HuggingFace Run it on your own data, free

Identifiers

Model ID

microsoft/trocr-large-printed

Feature URI

mixpeek://image_extractor@v1/microsoft_trocr_large_v1

Overview

TrOCR is an end-to-end text recognition model that uses a pre-trained image Transformer (DeiT) as the encoder and a pre-trained language model (RoBERTa) as the decoder. The large variant achieves state-of-the-art on printed text benchmarks.

On Mixpeek, TrOCR extracts readable text from images and video frames, making text-in-image content searchable through natural language queries.

Architecture

Encoder-decoder transformer: DeiT-Large (24 layers) as image encoder, RoBERTa-Large (24 layers) as text decoder. Pre-trained on large-scale synthetic printed text data, fine-tuned on SROIE and IAM datasets.

Mixpeek SDK Integration

// No extractor parameter takes a Hugging Face model id (checked against
// GET /v1/discovery/extractors, which returns 13), so trocr-large-printed runs
// on your side and the output is upserted through POST
// /v1/namespaces/{namespace_id}/documents/upsert. On Enterprise the other
// path is to upload the weights instead: POST /v1/namespaces/{id}/models
// accepts the huggingface format and a custom plugin loads them.
const res = await fetch(
  "https://api.mixpeek.com/v1/namespaces/ns_your_namespace/documents/upsert",
  {
    method: "POST",
    headers: {
      Authorization: "Bearer API_KEY",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      collection_id: "col_your_collection",
      documents: [
        {
          document_id: "asset-00412",
          // The model produces text, so it lands in payload. Give the
          // collection a text vector index and embed that text to make it
          // searchable rather than only filterable.
          payload: { extracted_text: modelOutput, source_key: "archive/2026/asset-00412" },
          vectors: { "text-embedding": embeddingOfModelOutput },
        },
      ],
    }),
  },
);

// Managed alternative, if this exact model is not the requirement:
// universal_extractor@v1 runs google/gemini-embedding-2
// (3072-d) over a bucket, with no inference of your own.