SOD-YOLO (Small Object Detection YOLO) builds upon the foundational YOLOv8 model to address the unique challenges of detecting small objects in complex backgrounds typical of UAV imagery. This repository contains the code for the model, including the novel enhancements we have introduced.

This repository contains implementation for Dmitrii I. Krasnov, Sergey N. Yarishev, Victoria A. Ryzhova, Todor S. Djamiykov paper "Improved YOLOv8 Network for Small Objects Detection"

🛰️ SAHI-powered satellite reconnaissance system for detecting military vehicles in high-resolution imagery. Uses YOLOv8 with sliding window inference to detect small objects (tanks, trucks, AFVs) that standard detection misses due to aggressive image downscaling. Processes 4000x4000 images via 512x512 overlapping tiles.

Code for "YOLOv8-SMOT: An Efficient and Robust Framework for Real-Time Small Object Tracking via Slice-Assisted Training and Adaptive Association"

Enhanced YOLOv8 for Small-Object Detection in Multiscale UAV Imagery: Innovations in Detection Accuracy and Efficiency

Real-time Varroa mite detection system using YOLOv8 and edge AI. Trained on 15,000+ images, it runs on Raspberry Pi 5 and Hailo-8L with 89.8% mAP@0.5. Includes a web dashboard for live results. Designed for precision beekeeping with an emphasis on small object detection and real-world deployment.

No description available

Object Detection for small objects using YOLOV8 and Slicing Aided Hyper Inference)

YOLOv8-SLG: A Small-Object Detection Model for Gobiocypris rarus larva in Laboratory Settings

The code for TVC 2025 paper "A Lightweight Enhanced YOLOv8 Algorithm for Detecting Small Objects in UAV Aerial Photography".

This is offcial repository for optical remote sensing with improved yolov8 to detect high resolution small object.

Evaluating YOLOv8 fine-tuning strategies for fine-grained playing card detection in real-world, small-object scenarios.

It is a small object detection based on improved YOLOv8 for airport scene images in haze weather.

MFA-YOLO: A Multi-Scale Fusion and Attention Mechanism Enhanced YOLOv8 for Small Object Detection in Remote Sensing Images

This project uses the YOLOv5 small and YOLOv8 small for the object detection and computer vision to read image , video , draw rectangle , and to put text label

Implementation and evaluation of object detectors (YOLOv1, Faster R-CNN, RetinaNet, YOLOv8, DETR) on the SeaDronesSee dataset, a collection of drone-captured images designed for small-object maritime detection.

A professional-grade aerial surveillance system using YOLOv8 and Unsupervised Machine Learning. Features high-resolution inference (1280px) for small object detection and an intelligent anomaly-based target locking system.

UAV aerial object detection is challenged by complex backgrounds, dense small targets, and frequent occlusions, while limited on-board computational resources make it difficult to deploy high-accuracy models in real time. To address these challenges, we propose AeroFineFusion-Det, a YOLOv8-based model for small-object detection .

A small project demonstrating traffic object detection and instance segmentation using Ultralytics YOLOv8 models. This repository contains a template script for running inference, and two pretrained model weight files (yolov8s-seg.pt and yolov8x-seg.pt).

echnical benchmark comparing YOLOv8 (2024) and YOLO26 (2026, NMS-Free) across three inference modes on a Tesla T4 GPU: Oriented Bounding Box (OBB) for aerial/satellite imagery, Standard Detection (COCO) for general-purpose video, and SAHI (Slicing Aided Hyper Inference) for small-object recall enhancement.

Optimized YOLOv8 small model for object detection and tracking

Small Project for Technical Interview, learn to use Google Colab and YOLOv8 to detect and track objects in real-time

Hybrid AI system using YOLOv8 + OpenCV for real-time detection of small hazardous objects (FOD) with motion detection and alert system.

Small YOLOv8 example detecting flying objects. Trained on few images, it can analyze videos in real time. For demo and experimentation.

A performance comparison of YOLOv8 Nano and Small models for real-time soccer analytics, demonstrating that the Small model achieves a 30.9% higher $mAP_{50}$ and superior detection of small objects like balls.

This study presents a groundbreaking approach to enhance the accuracy of the YOLOv8 model in object detection, focusing mainly on addressing the limitations of detecting objects in varied image types, particularly for small objects.

YOLOv8 Object Detection Project - Comparing YOLOv8 Nano vs Small models for accuracy and speed trade-offs. Includes training with COCO8 dataset, performance metrics, and internship presentation-ready comparison table.

A hybrid object detection approach combining YOLOv8 and Markov Random Fields for detecting small, distant birds in wildlife camera trap images.

Python project that uses YOLOv8 for real-time object detection and speaks out what it sees using Text-to-Speech. This can be a small step toward creating assistive technology for the visually impaired. 🔍 YOLOv8 detects people, objects, etc. 🗣️ pyttsx3 gives them a voice.

A YOLOv8-based aerial small-object detection system for identifying and counting people in drone imagery, optimized for high-resolution training and improved localization accuracy.