Detect handwritten words (neural network based).

꼼꼼한 딥러닝 논문 리뷰와 코드 실습

Official implementation for ICDAR 2021 paper "Handwritten Mathematical Expression Recognition with Bidirectionally Trained Transformer"

Document Text Recognition (DocTR) made seamless, high-performing & accessible to anyone using Deep Learning for OCR-related tasks.

Reformer, the efficient Transformer, in Pytorch

[CVIU 2019] ASSD learns to highlight useful regions on the feature maps while suppressing the irrelevant information, thereby providing reliable guidance for object detection.

Minimal keyword extraction with BERT

[WACV2021] Oriented Object Detection in Aerial Images with Box Boundary-Aware Vectors

More readable and flexible yolov5 with more backbone(resnet, shufflenet, moblienet, efficientnet, hrnet) and (cbam)

ISTR: End-to-End Instance Segmentation with Transformers (https://arxiv.org/abs/2105.00637)

[kaggle] 3rd place solution 🌿

YOLOv5 in PyTorch > ONNX > CoreML > TFLite

Self-supervised vIsion Transformer (SiT)

Show, Attend, and Tell | a PyTorch Tutorial to Image Captioning

An Arabic text localization and recognition system supporting multiple architectures for the recognition of text.