Save Birdlife: A Novel Deep Learning method for Detecting and Automatically Diversifying Birds from the Window Glass

摘要

According to a BBC news report: Window strikes are among the top three human-related causes of bird deaths, along with cats and habitat destruction. Nearly one billion birds die each year in the United States due to collisions with windows, and research reports on this case show that 54-76 percent of window collisions are mortal. Because Sometimes, birds do not perceive windows as a barrier. Instead, reflections in glass look to them as open areas, so they fly directly into the glass-window. The absence of depth perception may contribute to birds" false belief that the reflection they have seen is a bird belonging to their habitat. So, it is our responsibility to save birds. We came up with a creative project by which we can reduce the death rate of birds. We are proposing a deep learning method by using an object detection technique. We came up with a new solution using which we can divert the birds through wooden curtains(blind). They operate according to the instructions that are passed from the Arduino. When the model detects the bird that is coming close towards the window glass, the window blind will automatically shut down and even if the bird hits, it won’t collide instantly as the momentum will decrease gradually. The window blind will automatically open when there is no bird. It is a useful technique by which we can divert the birds. This technique has multiple credits like we are saving birds with a practical and cost-efficient approach.