​Summary

​This workshop is designed for software developers, AI practitioners, students, and engineers interested in deploying computer vision models on edge devices. Participants should have a basic understanding of neural networks and common deep learning concepts such as convolutions, activation functions, model parameters, and data flow through a network. Prior experience with deploying AI models to embedded or edge hardware is not required.

​The workshop focuses on the practical challenges of bringing computer vision models from development environments into real-world applications. Participants will gain hands-on experience with modern edge AI workflows and learn how deployment decisions impact performance, accuracy, and resource utilization.

​Syllabus

​The workshop will focus on the practical deployment of computer vision models to Luxonis edge AI devices. Starting from a trained model, participants will follow the complete deployment workflow and learn how to optimize models for efficient execution on resource-constrained hardware.

​Topics include model conversion, quantization, calibration, operator compatibility, graph and operator fusion, performance profiling, and deployment validation. Participants will learn how to evaluate the trade-offs between accuracy, latency, throughput, and power consumption, as well as how to systematically compare deployed models against their original training results.

​By the end of the workshop, attendees will have a clear understanding of the key considerations and best practices involved in deploying production-ready computer vision applications on edge devices.

​Instructors

​All three instructors are members of the Luxonis team, a company specializing in edge AI and computer vision solutions. Matija Teršek is Luxonis' CTO, with extensive experience in computer vision systems and AI hardware. Klemen Škrlj leads AI development, focusing on the design, training, optimization, and deployment of computer vision models. Aljaž Konec is part of the AI team and brings expertise in software and firmware development, system integration, and deploying AI workloads on edge devices. Together, they combine practical industry experience across the full computer vision deployment pipeline.