This short course shows you how to build reliable vision datasets and configure detection models with confidence. You’ll learn how to run a quality-controlled annotation process, review bounding boxes, coach annotators, and check dataset consistency using IoU-based audits. You’ll also explore how to analyze object sizes with clustering to generate anchor box parameters for models like YOLOv8. Through compact videos, guided readings, and hands-on exercises, you’ll practice using tools such as CVAT and Python notebooks to complete tasks common in production vision teams. By the end, you’ll be able to create a clean bounding-box dataset and use real measurements to tune model anchors—skills that support robust, scalable computer-vision pipelines.

Build & Evaluate Real-Time Object Detectors is an intermediate hands-on course for ML engineers who need to deploy fast, accurate object detectors under real-world constraints. When accuracy falls short of KPIs, or FPS drops below target, you need the skills to diagnose metrics, recommend improvements, and evaluate whether a real-time pipeline meets requirements. You'll learn how to compute and interpret detection metrics like mAP and APsmall, identify causes of underperformance, and propose targeted improvements. Then you'll analyze a complete real-time detection pipeline using models like YOLOv8 and trackers like DeepSORT, and evaluate it against throughput requirements such as 25 FPS at 720p. Through short videos, practical readings, analysis-based labs, and a final graded assessment, you will develop the skills to evaluate detectors, recommend optimizations, and assess whether solutions meet real-time demands.

This short course helps you improve segmentation models when classes are heavily imbalanced and predictions show recurring errors. You will learn how to apply class-balancing strategies such as focal-dice hybrid loss and sampling adjustments on medical or industrial datasets where foreground pixels may be extremely rare. You will also learn how to analyze predicted masks using region measurements to spot over-segmentation, under-segmentation, and shape-specific failures. Through concise videos, hands-on activities, and reflective checkpoints with Coach, you will practice improving recall, inspecting connected components, and building simple error logs that uncover patterns. By the end, you will have a repeatable approach for balancing datasets and diagnosing mask-level errors in production-ready segmentation workflows.