It is done on Python using COCO ana VOC datasets that get the photos and determines the objects in that image
according to a given ground truth
It's done using 3 different models ( Retina-ResNet , SSD 300 VGG 16, Faster RNN)