This provides a benchmark for cyclist's orientation detection, "CIMAT-Cyclist" with bounding box based labels according to eight different classes depending on the orientation. Which contains 11, 103 images, of which 6,605 images were collected in approximately 450 videos and images taken from sports events and the streets of the state of Zacatecas, Mexico, while 4,498 additional images were obtained from the web in pages such as pixabay, pexels, freephotos, among others. "CIMAT-Cyclist" provide 20,229 instances over 11,103 cyclist's images, where 80% of the images were split for the training set and 20% for the test set.
2 PAPERS • 1 BENCHMARK
Over 20,000 annotated synthetic images and web-scraped images of bicyclists with bounding box annotations in Pascal VOC format.
1 PAPER • 1 BENCHMARK