Handwriting OCR Data On Board of 8 Languages | 9,574 Images | Computer Vision Data

9,574 Images – Handwriting OCR Data On Board of 8 Languages. The language distribution includes English, Spanish and Portuguese, etc. The data diversity includes multiple collecting scenes, multiple text carriers, multiple languages, multiple photographic angles. The collecting angeles are looking up angle, eye-level angle and looking down angle. In terms of annotation, row-level quadrilateral (polygon) annotation and content transcription are performed on the text. The dataset can be used for tasks such as handwriting OCR. Data size 9,574 images, 243,240 bounding boxes Language distribution English, Spanish, Portuguese, French, German, Japanese, Italian and Dutch Collecting environment black boards, white boards, green boards Device cellphone Photographic angle eye-level angle, looking down angle, looking up angle Data format the image data format is .jpg and other common image formats, the annotation file data format is.json Annotation content line-level quadrilateral (polygon) bounding box annotation and transcription for the texts Accuracy rate the error bound of each vertex of quadrilateral bounding box is within 5 pixels, which is a qualified annotation, the accuracy of bounding boxes is not less than 95%; the texts transcription accuracy is not less than 95%