Kashmiri OCR resources
Datasets and models for reading printed and handwritten Kashmiri from images, most of them for the Perso-Arabic Nastaliq script.
- 17
- Resources2024–2026
- 7
- Datasets
- 3
- Models
- 5
- Papers
Filter this task in the full hub · Guide: Kashmiri text technology · Kashmiri OCR: Where Things Stand for Printed Text and Manuscripts
Datasets 7
BhartiOCR
Synthetic OCR and document benchmark: 13.25M image–text pairs across 23 South Asian languages and 12 writing systems, as WebDataset shards. Includes Kashmiri.
Koshur Pixel — 613K Synthetic OCR Pairs
613,078 synthetic image-text pairs for Kashmiri OCR, spanning multiple fonts and granularities from individual words to full pages, with 25+ augmentation strategies simulating real document degradation.
Koshur-OCR-Synth
Large synthetic Kashmiri OCR dataset: clean text-line images rendered in Kashmiri Nastaliq and Naskh fonts, each paired with its exact Unicode transcription.
Sarvam Indic OCR Bench
OCR benchmark of 6,909 curated text blocks from real document pages (19th century to present: textbooks, newspapers, magazines) in the 22 scheduled languages plus English, including Kashmiri.
Handwritten Kashmiri Text Dataset
Images of handwritten Kashmiri text (about 236 MB) shared on Kaggle for handwriting recognition. Non-commercial licence.
40K Kashmiri Text-Image Dataset (OCR)
About 40,000 Kashmiri text images paired with their transcriptions, for training and testing OCR models. Released by HNM; the related 600K-KS-OCR corpus is described in its own paper.
Kashmiri Sample Text Recognition (OCR)
5,000 samples of Kashmiri text images paired with text labels for OCR model training and testing.
Models 3
Koshur-OCR (KashlabsAI)
An OCR model for Kashmiri script from KashlabsAI.
KoshurOCR-CRNN Baseline
Two-stage Kashmiri OCR pipeline: a YOLOv8 detector segments words and a ResNet-34 + BiLSTM + CTC network reads them, to avoid the drift of single-pass Nastaliq line recognition.
KashViT — Kashmiri Word OCR
Two Vision Transformer models for recognising single Kashmiri words, with training, evaluation and inference scripts and a small web app to compare them.
Papers 5
600K-KS-OCR: A Synthetic Corpus for Kashmiri Script Recognition
Introduces a 602K-word synthetic OCR dataset for Kashmiri in three typefaces (Naskh, Nastaleeq, Nakash). Evaluates CRNN and TrOCR architectures for Perso-Arabic script.
A Baseline Optical Character Recognition Framework for Printed Kashmiri Nastaliq Text Using Deep Learning
Baseline OCR for printed Kashmiri Nastaliq: the open-source Kraken engine, transferred from a pre-trained Arabic model and fine-tuned on synthetically rendered Kashmiri lines. Reports 68% character accuracy on real-world text.
Koshur Pixel: A Large-Scale Synthetic OCR Dataset for Kashmiri
First large-scale synthetic OCR dataset for Kashmiri: 613,078 image-text pairs generated from the KS-PRET-5M corpus using the SynthOCR-Gen framework. Spans multiple fonts and granularities from single words to full-page documents, with 25+ augmentation strategies emulating real-world document degradation.
Multilingual OCR for Cursive Scripts Using Swin Transformers
Handwritten-text recognition with Swin Transformers for cursive scripts, including Kashmiri.
SynthOCR-Gen: A Synthetic OCR Dataset Generator for Low-Resource Languages
The generation framework behind Koshur Pixel. Produces synthetic OCR training data for low-resource scripts, breaking the annotation bottleneck that blocks OCR development for languages without large labelled document collections.