About Me

I’m John C. Femiani, the Professor of Computer Science at Miami University in Oxford, OH. My research interests are in artificial intelligence (AI), machine learning (ML), computer vision,computer graphics, and remote sensing with a focus on using advanced learning models to extract structured, meaningful representations from complex datasets such as images and video.

Research Interests and Work

My work lies at the intersection of AI, machine learning, and computer vision, where I develop methods to process and understand large, multimodal datasets. I focus on algorithms that extract meaningful patterns from minimal training data, particularly in applications involving remote sensing, scene reconstruction, and generative models. My research aims to push the boundaries of unsupervised and self-supervised learning, enabling models to perform tasks with less reliance on large, labeled datasets.

Core Areas of Interest:

• Artificial Intelligence and Machine Learning: Developing AI systems that generalize with fewer examples, utilizing techniques such as one-shot and zero-shot learning. I aim to create models that learn efficiently from limited data.
• Computer Vision: From detecting structures like buildings and roads in satellite imagery to generating 3D models from 2D data, my work addresses how machines can interpret and interact with visual information.
• Generative Models: My recent work involves leveraging GANs and transformers for scene generation and image manipulation, including projects that enable interactive, real-time editing of images using minimal user input.

These efforts tie into a broader goal: creating models that can operate effectively with less supervision and more autonomy, mimicking human cognitive processes in understanding complex scenes.

Academic Background

My academic path began at Arizona State University (ASU), where I initially pursued a degree in fine arts. However, my interests quickly shifted to the technical aspects of 3D modeling and computer graphics, and I transitioned to computer science.

I later joined the Partnership for Research in Spatial Modeling (PRISM), where my work focused on surface parameterization and the analysis of 3D geometric structures. My dissertation revolved around the problem of document image understanding, specifically the segmentation of handwritten versus typeset marks.

Research Milestones:

• Surface Parameterization: Early work in 3D surface analysis, developing mathematical models for representing complex surfaces.
• Document Image Understanding: Research on segmenting and classifying mixed-content documents using optimization techniques.
• Remote Sensing Applications: Developing algorithms for extracting urban features (e.g., roads, rooftops) from satellite imagery, contributing to 3D reconstruction and urban modeling.
• Generative AI Research: Focused on multimodal generative models, initially on GANs and now diffusion models. Current work explores the use of large language models (LLMs) to enhance computer vision tasks, enabling richer semantic understanding and control in visual data generation.

Advancing AI and Machine Learning

With the rise of deep learning, I have expanded my work to focus on how neural networks can be applied to problems in image classification, material recognition, and 3D scene modeling. I am particularly interested in inverting generative models like GANs to allow interactive editing of scenes and objects.

More recently, I’ve been exploring transformer-based architectures and how they can integrate language models with visual data for tasks such as zero-shot image classification. This fusion of vision and language aims to improve the interpretability and usability of AI systems in real-world applications.

Personal Interests

While much of my time is devoted to research, I also enjoy weightlifting, fishing, and fossil-hunting. These activities provide a balance between my academic pursuits and personal interests, offering me a way to stay connected with the natural world and physical well-being.

For collaboration or further inquiries, you can reach me at femianjc@miamioh.edu.