--- license: mit language: - en base_model: - Ultralytics/YOLO26 tags: - garbage - detection - recycling - plastic - paper - cardboard - glass - shoes - battery - imagerecognition - recognition - environment - trash --- # YOLO26 Class Based - Garbage & Recycling Object Detection Model

val_batch2_labels

## Model Summary This model is a fine-tuned version of **Ultralytics/YOLO26 Classification** optimized for real-time detection of trash, garbage, and common recyclable materials. Developed to support waste management systems, automated recycling sorting lines, and environmental cleanliness monitoring applications, this model balances exceptional inference speed with high precision across distinct waste categories. ## Model Description - **Developed by:** [barbaramartina] - **Model Type:** Object Classification - **Base Model:** `Ultralytics/YOLO26 Class` - **Language(s):** English (`en`) - **License:** MIT License ### Supported Detection Classes The model has been explicitly trained to identify and categorize the following objects: 1. **Plastic** (Bottles, containers, wrappers) 2. **Paper** (Sheets, newspapers, flyers) 3. **Cardboard** (Boxes, shipping packaging) 4. **Glass** (Bottles, jars) 5. **Shoes** (Discarded footwear) 6. **Battery** (E-waste items requiring special hazardous handling) 7. **General Trash / Garbage** (Non-recyclable domestic or industrial waste) --- ## Intended Uses & Limitations ### Intended Uses - **Smart Recycling Bins:** Integrating with IoT devices to automatically classify thrown-away items and guide consumers toward proper sorting. - **Hazardous Waste Filtration:** Flagging critical items like batteries out of standard trash streams to prevent fires or environmental contamination. - **Urban Cleanliness Automation:** Deploying on street sweepers, drones, or stationary city cameras to map litter hotspots. ### Limitations & Biases - **Lighting Conditions:** Performance may degrade in extremely low-light or heavily shadowed environments. - **Deformed Objects:** Heavily crushed, burnt, or shredded materials may experience lower confidence scores compared to intact objects. - **Occlusion:** Multiple stacked items or items hidden inside translucent bags may result in partial or missed detections. --- ## How to Get Started You can easily integrate this model into your project using the `ultralytics` Python framework. ### Installation ```python pip install ultralytics from ultralytics import YOLO # 1. Load the model # Replace the following with the path to the model model = YOLO("your-username/your-model-id") # 2. Run inference on an image or video source results = model("path/to/waste_image.jpg", conf=0.25) # 3. Check the results (I suggest using jupyter notebook) ```