Workshop designed for the next generation of Thai AI engineers. Learn how modern object detection systems are optimized, compressed, accelerated, and deployed under real industrial constraints.
Unlike typical tutorials focused only on benchmark accuracy, this workshop explores the hidden engineering trade-offs behind deploying scalable object detection systems in production. Students will experience how real AI products are engineered under computational, memory, and deployment constraints.
Understand how detection systems balance inference speed, throughput, and accuracy in real-time environments.
Learn pruning, quantization, and lightweight architecture design for efficient deployment.
Explore production-scale pipelines, multi-camera inference, and GPU scheduling strategies.
Six hands-on labs covering optimization techniques, scalable deployment, and industrial computer vision systems.
Benchmark latency, FLOPs, FPS, and mAP across different YOLO variants.
Compress object detection models while preserving performance.
Deploy INT8 and FP16 optimized detectors for faster inference.
Transfer knowledge from large teacher models to compact student networks.
Design scalable multi-scale detection heads and efficient backbones.
Build scalable AI pipelines for production-scale video analytics systems.
This workshop goes beyond academic toy examples. Students will experience real deployment constraints, engineering trade-offs, and industrial AI thinking.
The future of AI engineering is not only about building larger models, but building systems that are deployable, scalable, efficient, and practical in the real world.
Complete lecture slides for AI optimization, deployment strategies, and scalable object detection systems.
Interactive workshop notebook for profiling, pruning, quantization, and deployment experiments.
Official workshop repository containing lecture materials, notebooks, and educational resources.