Overview / Usage

To assess the impact of climate change on Earth's flora and fauna, it is vital to quantify how human activities such as logging, mining, and agriculture are impacting our protected natural areas. Researchers in Mexico have created the VIGIA project, which aims to build a system for autonomous surveillance of protected areas. The first step in such an effort is the ability to recognize the vegetation inside the protected areas. The task is to come up with an algorithm that can identify a specific type of cactus in aerial imagery.

Methodology / Approach

• The dataset labels are very well distributed and do not suffer from any skew
• The dataset has a good number of samples
• The colors of the images in the dataset are quite varied and there are various patterns amongst them
• Deep learning just felt like the ideal solution to me for this
• Created data augmentation pipelines for better generalization
• Created a separate validation set
• Used a pretrained ResNet34 architecture
• Used one cycle policy for training the model
• Fine tuned it and also used discriminative learning rates
• Averaged out the predictions for better precision

Technologies Used

• Python
• fastai, numpy, PyTorch
• Kaggle Kernels

Repository

https://github.com/sayakpaul/Aerial-Cactus-Identification