GradCAM for the BreaKHis Dataset
An NBDev package for fine-tuning ResNets to visualize gradient-weighted class activation for the BreaKHis dataset.
medical-imaging computer-vision transfer-learning
Objectives & Highlights

• Use transfer learning and cyclic learning rate schedules to quickly obtain an accurate classifier for the task • Use gradient-weighted class activation maps to understand how the classifier is performing • Use NBDev to wrap the code from notebooks into a package with documentation

Don't forget to add the tag @dthiagarajan in your comments.

My name is Dilip Thiagarajan - I’m an incoming AI Engineer at Paige, and was formerly a software engineer and intern at FB, on their Business Integrity ML, Computer Vision, and Ads Content Understanding teams (from most recent to least recent). I hold a Masters of Engineering in CS, with a specialization in computer vision and machine learning, from Cornell University. I’m interested broadly in the application of computer vision in medical applications, as well as Bayesian machine learning, and statistical learning.
Share this project
Similar projects
Brain Tumor Segmentation BRaTS 18
Segmentation of gliomas in pre-operative MRI scans. Use the provided clinically-acquired training data to produce segmentation labels.