The this project aims to apply Semantic Segmentation, a deep learning technique, to process and analyze images captured by drones. Semantic Segmentation is a high-level task that focuses on partitioning an image into different segments. It's about understanding an image at a pixel level i.e., assigning each pixel in the image an object class.
Semantic segmentation has a wide variety of applications and is essential in scenarios where precision is necessary. For example, in drone or satellite imagery, it can be used for land mapping, agricultural field detection, and urban planning among many other applications.
This project utilizes the PyTorch library, a popular open-source machine learning library for Python known for its flexibility and efficiency, to construct the deep learning models.
The dataset used in this project consists of aerial images captured by drones. The model is trained to understand these images at a pixel level, effectively learning to differentiate between various geographical and man-made features within the images. The resultant segmented images provide a detailed perspective of the landscape, which can be further utilized for different analyses and planning.