George Kalitsios's Projects
3D Convolutions for Video
Computer Vision Nanodegree - Automatic Image Captioning Project [ Python ] : Combine CNN and RNN knowledge to build a deep learning model that produces captions given an input image.
Deep Reinforcement learning
Master Thesis Algorithms
Unlike traditional computer vision techniques, DETR approaches object detection as a direct set prediction problem. It consists of a set-based global loss, which forces unique predictions via bipartite matching, and a Transformer encoder-decoder architecture. Given a fixed small set of learned object queries, DETR reasons about the relations of the objects and the global image context to directly output the final set of predictions in parallel. Due to this parallel nature, DETR is very fast and efficient.
Computer Vision Nanodegree - Facial Keypoint Detection Project [ Python ] : Use image processing techniques and deep learning techniques to detect faces in an image and find facial keypoints, such as the position of the eyes, nose, and mouth on a face.
Computer Vision Nanodegree - Landmark Detection and Tracking (SLAM) Project [ Python ] : Use feature detection and keypoint descriptors to build a map of the environment with SLAM (simultaneous localization and mapping).
Itβs time to get rid of the black boxes and cultivate trust in Machine Learning
Mask R-CNN for object detection and instance segmentation on Keras and TensorFlow
AI & Games Project
The large-scale Synthetic Power Lines (SPL) dataset was generated using unity, one of the most popular gaming engines in the world, and unity's perception package. The introduced dataset contains aerial images of power lines in every possible combination of different locations, lighting conditions, tilts, angles, FOVs.
Vision-Enhanced System for Human-Robot Disassembly Factory Cells: Introducing A New Screw Dataset