In this project, our team is trying to classify image data of scenes we might find in the world around us today. Our goal is to use the power of Computer Vision to see if we can develop models to be able to predict images we receive into one of six categories: Building, Forest, Glacier, Mountain, Sea, and Street. As such, this is a clear classification task. This project is important because it will allow our team to acquire real-world experience in creating Deep Learning models using the latest data science tools. Gaining this valuable knowledge will equip us to add value to the image classification domain in the future, which is highly utilized for so many critical tasks. Popular applications of this field include self-driving cars, medical imagery, and face recognition. The most complex solutions start with the simplest building blocks, the latter being the aim for us to master in this project by building upon the exposure given to us in class.
We utilized documentation on a variety of libraries in order to gain a better understanding of Deep Learning, Image Classification, and Image Augmentation. This was done by looking primarily into three important libraries for these tasks: Keras, TensorFlow, and OpenCV. These sources allowed us to see how to use the latest technology to solve our problems, and our project was greatly inspired by the tutorials and samples available on their sites. Likewise, learning about the challenges with image classification such as noise, angle of objects, image quality, and scale of objects gave us more insight into how this may affect the images that we will be testing our models with. Using research papers that involved image classification in order to solve problems, such as identifying plant species with specific diseases, was a great reference for learning how to structure our own analysis and communicate with the audience. Moreover, using blogs in order to peek into the types of problems and approaches taken when structuring an analysis involving pixel data have been reflected in our project.