Annotated Bangla Handwritten Word Images for Character Detection (HandwrittenWordsB)
Dataset Information
Keywords
bangla writing, handwritten words, characters annotated, train YOLO
Description
We collected a dataset comprising of 300 handwritten word images.
Newspaper articles containing a total of 10,278 words were collected from Prothom Alo to find the word length distribution of commonly used Bangla words. From the articles, a list of unique words were filtered out and their lengths were analyzed. This distribution was followed when generating the word list for data collection. 20 sets containing 15 words were created to distribute among 20 participants; each participant was given a unique set. The dataset contains actual handwritten text noises like ink blots, smudges, and irregularities in writing style; this guarantees that the feature learning is carried out in a real-world environment.
File | Type | Size | Downloads | Description |
---|---|---|---|---|
Handwritten Words.v1i.yolov8.zip | data | (12 MB) | 1 |