Small Object Detection in Aerial Imagery
YOLOv8 variants for tiny birds in drone frames—benchmarked for real-world ecology and surveillance use cases.
Problem & industry context
Aerial imagery makes objects appear small and dense; generic detectors tuned on COCO underperform on birds, drones, and wildlife monitoring. Researchers compare anchor-free detectors (YOLOv8 family) on domain-specific datasets and report precision-recall tradeoffs per model size (nano vs medium vs extra-large).
Insight
Small-object detection is a scale and resolution problem before it is an architecture fad. Choosing yolov8n vs yolov8x is a deployment decision (edge drone vs server batch), not only a leaderboard exercise. Cross-domain papers should document failure modes—occlusion, motion blur, class imbalance.
What I built
Benchmarked YOLOv8n, yolov8m, and yolov8x for small bird detection in aerial imagery. Published at AIP ICRAIC 2024 (Scopus-indexed). Highlighted practical implications for avian ecology and surveillance pipelines.
Technical approach
Stack and tooling for this work: Python, YOLOv8, Deep Learning, Object Detection. Topics covered: YOLOv8, Computer Vision, Aerial Imagery, Research.
Topics
Related case studies
AI-Driven Exam Paper Quality Assessment
CHARUSAT Research · 2023 – 2024
NLP that scores question-paper quality against Bloom's taxonomy—98.04% accuracy, Springer ICTIS 2024.
FinTech Data Pipeline at Scale
Digiflux Technologies · 2024 – Present
How 50+ serverless scrapers keep market and compliance data flowing for millions of users.
Conversational AI for Meal Subscriptions
Digiflux Technologies · 2024 – Present
LangChain + LLMs powering subscription support for 125K+ customers with 90%+ intent accuracy.