Converting CROHME dataset for Online-handwritting recognition to Offline-handwritting recognition.
-
Updated
Aug 24, 2026 - Python
Converting CROHME dataset for Online-handwritting recognition to Offline-handwritting recognition.
A modified extractor for the CROHME handwritten math symbols dataset.
data synthesis for simulation of pen-based interaction
MathSnap AI — An end-to-end handwritten mathematical expression recognition (HMER) system powered by Mini-CoMER, a compact encoder-decoder architecture (6.39M params) with DenseNet encoder, Transformer decoder, and Attention Refinement Module (ARM). Trained on CROHME 2013/2016/2019 datasets.
Scaling Up Image-to-LaTeX Performance: Sumen An End-to-End Transformer Model With Large Dataset
A Faster R-CNN object detection model for recognizing handwritten mathematical symbols on whiteboards. Built as the core detection engine for an end-to-end math expression visual retrieval pipeline.
SSL Methods integrated into the official implementation of the ECCV"22 Paper "CoMER: Modeling Coverage for Transformer-based Handwritten Mathematical Expression Recognition"
Handwritten math-equation recognizer: ResNet-18 encoder + Transformer decoder, image-to-LaTeX on the CROHME 2013 benchmark (16.5% exact match / 50% token accuracy on the official test set).
To associate your repository with the crohme topic, visit your repo's landing page and select "manage topics."