
Trash Detection System 🗑️🔍
Project Type: Group ProjectTechnologies Used: Python, TensorFlow/Keras, OpenCV, CNN, YOLO, Pandas, Matplotlib, SeabornDataset: Custom Labeled Trash Images
Overview
This project focused on developing an object detection system to classify and identify trash in images. Our team built and compared two deep learning models: a Convolutional Neural Network (CNN) and YOLO (You Only Look Once). The goal was to determine the best-performing model in terms of accuracy and efficiency. I was responsible for developing and optimizing the CNN model.
Key Contributions
✅ Model Development: Implemented a CNN model using TensorFlow/Keras to detect and classify trash items.
✅ Data Preprocessing: Labeled and augmented image data, applied resizing, and normalized pixel values.
✅ Training & Optimization: Experimented with different architectures, dropout rates, and batch sizes to improve model performance.
✅ Evaluation & Comparison: Compared CNN with YOLO based on accuracy, precision-recall, and inference speed.
✅ Visualization: Created bounding box visualizations and accuracy/loss plots to analyze model performance
🔗 Interactivity & Links
📌 GitHub Repository: (https://github.com/mqarmout/TACO_object_detection.git)
📌Live Demo:( http://35.224.185.15:5000/)
Would you like to tweak any details or add more metrics?