Reducing Frequent Killing of Elephants in Train Collisions Using Machine Learning
Abstract
Elephant-human conflicts are a significant issue causing death and injury to elephants and humans. Elephants raiding human plants, unexpected humanelephant encounters as elephants cross their paths, elephants crossing railway lines are the major problems for these injuries. According to the report tabled in Department of Wild Life Conservation 279 elephant’s deaths have been reported in the last year (2018). The proposed system helps to identify elephants in day time as well as night time. This paper presents design and implementation of Deep Learning and PIR sensor-based solution to detect elephants in day time as well as night time and send a message to the user with the correct distance between elephants and train. In the implementation of deep learning model transfer learning has been applied with InceptionV3 model. The deep learning model gained 99.13% of training accuracy, 98.98% of validation accuracy and 98.86% of test accuracy with 0.0001 learning rate, 0.9 momentum and 4300 training images
Collections
- Computing [68]