← Back to Projects

BEESAFE

Edge AI Computer Vision for Varroa Mite Detection

Real-time computer vision system deployed on edge devices to detect Varroa mites on honeybees, demonstrating a complete machine learning lifecycle from data acquisition to hardware-constrained deployment.

Technical Pipeline

1

Data Acquisition

Dataset sourced and annotated using Roboflow's labeling tools

Roboflow
2

Model Training

Deep learning model trained on labeled honeybee images

Python
3

Optimization

Model converted to ONNX format for cross-platform compatibility

ONNX
4

Edge Deployment

Converted to TensorFlow Lite for mobile/edge devices

TFLite

Key Features & Impact

Edge Intelligence

Optimized for resource-constrained environments, enabling real-time inference on mobile devices without cloud dependency.

Agritech Application

Addresses a critical problem in beekeeping by providing early Varroa mite detection, helping protect bee colonies and agricultural ecosystems.

Technology Stack

Python TensorFlow Lite ONNX Roboflow Edge AI Computer Vision