Back
Under construction
- Id
- yolov8-pipeline
- Title
- YOLOv8 Model Training Pipeline
- Summary
- Automated labeling + retraining pipeline using CVAT, Jenkins, and YOLOv8 to iterate models per room.
- Tags
- ["Computer Vision","ML pipeline","GCP pub/sub","Jenkins"]
- Description
Dog Tracking Project
- Built an end-to-end computer vision system using Google Cloud Pub/Sub and the Nest SDM API to stream and process 1000s real-time camera events daily for indoor dog tracking.
- Trained and refined a custom YOLOv8 model, improving accuracy from 30% to 90%.
- Integrated with Label Studio for easy labelling and automated model retraining with Jenkins pipelines, enabling accurate pet monitoring and automated model improvement.