Elephant Poaching Risk Monitor

ELEPHANT POACHING RISK MONITOR

This project is mainly made for preventing Elephant poaching by monitoring and detecting the presence of human beings in the area of Elephants. The cameras and other sensors will always monitor the area where the Elephants are more found and always detect the presence of humans by using effective Machine Learning(ML) models. Once if humans are detected in the areas near Elephants then a notification message with the captured image and location will be sent to the wildlife rangers.

The classification and detection of elephants and humans are done with the help of a Machine Learning(ML) model. Machine Learning(ML) model is mainly written using the Python programming language and TensorFlow framework.

Edge Impulse Studio is used for creating an efficient ML model to classify and detect the presence of humans in the locations. Using the Edge Impulse Studio the model can be easily converted to a Lite model that can be also used on Android applications with the help of TensorFlow Lite. And also it helps to interface and deploy the model with any hardware components like Arduino, Xilinx, Raspberry, OpenMV, Web Assembly, and also on the web application by converting it into supporting libraries. Version control and Deployment are also very easy using Edge Impulse Studio.

The image classification model has been trained and developed using Edge Impulse Studio. The various steps involved in the Edge Impulse Studio for building an efficient ML model includes:

• Devices Edge
• Data Acquisition
• Impulse Design
• Retrain Model
• Live Classification
• Model Testing
• Versioning
• Deployment

The given below are the screenshots of the image classification development done on the Edge Impulse Studio platform:

Devices Edge

You can connect any of your devices to your Machine Learning(ML) model wireless or also by wired using the Edge Impulse Studio. The process connection of devices is given in below.

• Select the type of device which you like to connect to your model.

If you are connecting smart devices then you can connect using a QR code.

Data Acquisition

This is the major area where we are going to provide the data for the creation of the ML model.

Impulse Design

The below given is the screenshot of the Transfer Learning process inside the impulse design which is carried out with:

• Epoch Rate: 35 epochs
• Learning Rate: 0.005
• Minimum Confidence Rating: 0.80

Retraining Model

The Below given screenshot contains the retraining of the image classification model by using two known parameters:

• Image
• Transfer Learning

Live Classification

The given below is the live image classification process developed on Edge Impulse which gives an image classification with accuracy of 1.00 (i.e 100%)

Versioning

Using Edge Impulse Studios we can perform version control.

Deployment

We can deploy our impulse into any device, and we can make our project working on any devices even without an internet connection, minimizes latency, and runs with minimum power consumption.

For deploying our impulse into other devices first, we have to create a library. Using Edge Impulse Studios we can create a library easily by converting our impulse into optimized source code.

Impulse Creation Successful

Once your Impulse Creation is successful, the screen will be shown like below screenshot: