Textual Data collection
Budget: $250 – $750 USD
We require a dataset consisting of approximately 17,000 images of incidental text captured in various indoor environments.
Example: Bedroom, Kitchen, dining room, living room, Garage, Office, Workplace office, Reception/waiting room, Library, Coffee shop, Restaurant, Shopping center, Hair salon/ Barbers, Cinema/Theatre, Leisure center/gym, Museum/ art gallery.
The images should contain high-density text, which includes multiple lines of text such as those found in books, printed documents, and notice boards with handwritten post-it notes. The text within the images must be legible to human annotators. The annotation requirements for the images include high-quality labels for OCR tasks, such as text transcripts and bounding boxes for paragraphs, words, and lines within the text. The dataset should exclude Personally Identifiable Information, copyrighted material, and offensive content.
We value working with experienced vendors in imagery collection who can help us design a dataset that aligns with the specified requirements. It is important to note that any purchased dataset will be released on a public website under an open-source access license, providing high-quality data for ML model development in the Conversational AI domain.
The document also provides terminology definitions related to scene text, high-density text, text type, text images, scene text images, incidental scene text images, indoor incidental scene text images, focused text images, and more.
The dataset's purpose is to support the development of OCR technologies for text detection and recognition tasks. The dataset will be used to train and test models for text detection and text recognition, which involve identifying and localizing text in scenes and analyzing the extracted features to make predictions about words or characters.
Example: Bedroom, Kitchen, dining room, living room, Garage, Office, Workplace office, Reception/waiting room, Library, Coffee shop, Restaurant, Shopping center, Hair salon/ Barbers, Cinema/Theatre, Leisure center/gym, Museum/ art gallery.
The images should contain high-density text, which includes multiple lines of text such as those found in books, printed documents, and notice boards with handwritten post-it notes. The text within the images must be legible to human annotators. The annotation requirements for the images include high-quality labels for OCR tasks, such as text transcripts and bounding boxes for paragraphs, words, and lines within the text. The dataset should exclude Personally Identifiable Information, copyrighted material, and offensive content.
We value working with experienced vendors in imagery collection who can help us design a dataset that aligns with the specified requirements. It is important to note that any purchased dataset will be released on a public website under an open-source access license, providing high-quality data for ML model development in the Conversational AI domain.
The document also provides terminology definitions related to scene text, high-density text, text type, text images, scene text images, incidental scene text images, indoor incidental scene text images, focused text images, and more.
The dataset's purpose is to support the development of OCR technologies for text detection and recognition tasks. The dataset will be used to train and test models for text detection and text recognition, which involve identifying and localizing text in scenes and analyzing the extracted features to make predictions about words or characters.