I'm looking for publically available synthetic 2D-images datasets with ground truth for benchmarking of machine vision algorithms. Which ones do you know?
I'm working on some background theory on how to create such datasets. I'm not focused on a particular algorithm so far. Playing around with some segmentation stuff at the moment.
Which Machine learning algorithm you are planning to apply on datasets ? What is your aim ? The aim of your project will decide the kind of data set you need.
I'm working one some background theory on how to create such datasets. I'm not focused on a particular algorithm so far. Playing around with some segmentation stuff at the moment.
May be DRIVE (Digital Retinal Images for Vessel Extraction) database is of your interest. You can find Data description here: http://www.computervisiononline.com/dataset/drive-digital-retinal-images-vessel-extraction and download the dataset from: http://www.isi.uu.nl/Research/Databases/DRIVE/
There is another dataset called STARE (STructured Analysis of the Retina) , you can find its description here: http://www.aldiri.info/RelatedLinks.aspx
Have a look a the PUT face database:
https://biometrics.cie.put.poznan.pl/index.php?option=com_content&view=article&id=4&Itemid=14&lang=en
It's a base of images of faces with different poses/expressions with additional ground truth annotation for the eyes/eyebrows/mouth/nose/face contours.
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing