Leopard detection is done through Custom Object Detection using YOLO. Darkflow simplifies the use of YOLO and makes custom object detection simple and easier. Step-by-step how to use Darkflow can be found here here.
IIITDM Jabalpur campus is located right beside Dumna Nature Reserve. Dumna Nature Reserve is home to many faunae. Out of all of the faunae, Leopards and Deer have frequented the campus by jumping over the boundary walls. These Leopards pose a threat to the other residents of the lush green campus.
To make an IoT device that would alert authorities and people nearby about the prresence of a leopard, Custom Object Detection is used.
Real-time Custom Object Detection has been a major feild of research in Image Processing. There are pre-trained models of objects available, but there is none for a leopard. Using Darkflow, a model can be trained. The images used for training the model can be found here. The xml files of the images are available here. The trained ckpt files have also been made available here.
By running the test_live_cam.py file, the command prompt will show the results.
The proper steps are mentioned here.
This model for leopard detection has been made for StaySafe.
The whole system is implemented on Raspberry Pi with Pi NoIR Camera and GSM Module.
This project has been developed by Arnav Deep and Manjot Singh.
GNU GENERAL PUBLIC LICENSE Version 2
Arnav Deep © November 2019. All rights reserved.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent