Biometric Recognition of African Clawed Frogs

The dataset of African Claw Frogs consists of 1,647 images for 160 classes, with an average of 10.3 images per class. The frogs are housed in the University of Groningen, following animal welfare laws and approved by the relevant authorities. The frogs are divided into groups of ten to twenty individuals in aquatic tanks. To capture the images, the frogs are placed individually in transparent containers filled with water on a white surface. The setup takes into account the frogs' pigment changes by housing them in transparent boxes on a yellow surface for approximately 24 hours before photography. The pictures are taken using at least four different smartphones per frog, ensuring consistency in angle and distance from the camera to the container. The dataset includes images taken at different times of day, on different days, and with cameras of different smartphones to enhance the model's robustness. The dataset also provides information on the distribution of samples per frog and the number of photos taken per phone. This dataset has been used to develop an automatic pattern recognition system that is able to determine the individual frogs based on the analysis of their patterns extracted form the iamges. The developed algorithms are very effective and achieve more than 99% recognition rate.