The main aim is to develop a system which captures an image of the animals wandering on the roads within the specified range and compare it with the previously trained dataset with train ,test and validate differentation for the accuracy calculations.Intention is to alert the driver and avoid unfortuanate collision.
Around 3500 handpicked images of the 7 animals classifying into the 7 classes were collected for the training purpose and fed to the SSD MobileNet Algorithm.The images from the validte folder constitutuing about 10% of the total were used to adjust the hyperparameters like training speed.