Elephant Detection

Elephants are the largest terrestrial living species. They are herbivorous animals and require 100 kilograms to 200 kilograms of food and about 230 liters of water each day. Their home range can expand up to 11,000 square kilometers. Their ability to find food and water sources is gained from traditional knowledge learned over generations. This knowledge, which is important for survival, is lost if elder elephants of the herd perish.

Elephants are endangered due to many reasons. They can be killed by poachers for their tusks, or captured and tamed for social status and for the circus. Changes in environment such as global warming, rain patterns, deforestation, and mining can lead to degradation of their habitat, forcing these animals to move to different areas in search of food and water. This can cause conflicts with humans as elephants move into human settlements and farmlands. They can also run into electrical fences and traps.

To avoid life-threatening incidents, and for their conservation, monitoring the elephants and their movements is of high importance. It is easier to monitor the elephants using aerial imagery, as it does not require human intervention or disturbance in elephants' natural habitat. Elephant Detection using aerial imagery is more efficient when performed over vast areas. This deep learning model helps automate the task of detecting elephants from high-resolution aerial imagery.

Using the model
Follow the guide to use the model. Before using this model, ensure that the supported deep learning libraries are installed. For more details, check Deep Learning Libraries Installer for ArcGIS.

Fine-tuning the model
This model can be fine-tuned using the Train Deep Learning Model tool. Follow the guide to fine-tune this model.

Input
High resolution RGB imagery (3-13 centimeters spatial resolution).

Output
Feature class containing detected elephants.

Applicable geographies
The model is expected to work well with aerial imagery of southern African forests (South Africa, Botswana, and Namibia) or similar geographies.

Model architecture
This model uses the FasterRCNN model architecture implemented in ArcGIS API for Python.

Accuracy metrics
This model has an average precision score of 0.857 for elephant.

Training data
The model has been trained on the The Aerial Elephant Dataset.

Limitations

• This model works well only with high-resolution aerial imagery.
• This model is trained on imagery of African Bush Elephants. However, it detects all kinds of elephants and is species agnostic