Computer Vision News - August 2021

“ArkivDigital are going to share some more document images with us which were written during the 17th and 18th centuries. Reading, recognizing, and understanding the characters, digits, and sentences in there is going to prove even more difficult!” Practically, they see this work as being useful for anyone who is working on historical document analysis, particularly for Latin-based languages . Also, anyone looking to create stronger recognition and detection algorithms will benefit from the huge variety of writing styles available by combining DIDA with MNIST and USPS . The DIDA single digits dataset is publicly available for the research community at https://didadataset. github.io/DIDA/ . 31 DIDA and DIGITNET