Deep Learning Course on Anomaly Detection (Mars Version)

The anomaly/object detection course is about deep learning concepts with hands-on coding experience with a real-life case study detecting anomalies on the surface of MARS using OpenCV, Computer Vision, Pytorch, and the scikit-learn machine learning libraries.

Objective

The course objective is to deliver a hands-on, code-first experience with deep learning theories, models, and applications not as a state of the art skillset but also understanding and usefulness of applications and advancing AI for the betterment of humanities in the field of medicine, agriculture, and social good with computer vision and deep learning methods.

Our aim is to focus on deciding how real-world problems can be tackled with deep learning, determine the implementation of the best-suited model to solve the problem, and then visualize/justify the findings.

The application areas are not limited to only finding anomalies in crops, identifying affected areas with a particular disease in x-rays, facial recognition, and providing support to people with disabilities that can leverage computer vision/deep learning to its true essence with emphasis on the ethical perspective as well.

• Computer vision skillset
• Instructor-led online course
• Real-world practical assignment(s) leading to project
• Application in agriculture, Medicine, Facial Recognition, and assistive tool for people with special needs
• End to End solution from Archiving dataset to deployment phase

More details about the case study 

• Basic understanding of Computer vision / Image Processing
• Basic programming skills in python with PyTorch

The course content includes end-to-end solutions powered with AI from data collection to deployment of the project.

• Data Engineering: Data Collection, Storage, Image Processing and Labeling

• Getting started with Deep Learning:  Basic Models CNN, RNN

• Model Regularization and Optimization

• Application of Pre-trained Models

• GANs with variations
• Mask R-CNN
• U-NET
• Single Shot Multibox Detector (SSD)