Dog Poop Tracker

Find open-source links and Medium article at the bottom of this page.​​​​​​​

Meet Dog Poop Tracker, an innovative Python application that combines computer vision and machine learning to solve a common dog owner's challenge: locating and cleaning up dog poop in the backyard.

Using a camera, the application tracks my dog, Turing, with YOLOv8 and detects pooping behavior with a custom-trained MobileNetV2 model.

Dog Poop Tracker captures and saves dog poop images to help me to easily locate and clean poop

Following are example of poop/not poop training data:

Everything is open source! Below are the links to access the source code and dataset:
Dog Poop Tracker Source Code
Dog Poop Dataset
MobileNetV2 Training Notebook
Medium: Tracking Dog Poop in the Backyard: A Computer Vision Solution
Note: You can run the training script with the dataset directly on Kaggle using a free GPU, without downloading. The training process takes only 3 minutes on a Kaggle P100 GPU.
↑Back to Top