Recycla is an AI-powered smart bin that uses computer vision and embedded systems to classify and sort waste automatically. No labels to read, no confusion, no contamination.
Every day, thousands of recyclable items on university campuses end up in landfill because students sort incorrectly. Recycla exists to eliminate that uncertainty entirely.
We built a system that sees what you throw away, understands what it is, and puts it in the right place. No guessing. No signs. No guilt. Just accurate, automated sorting powered by a neural network trained specifically for campus waste.
Is a coffee cup recyclable? What about the lid? The sleeve? Rules change between municipalities, between buildings, between bins. Students default to guessing.
A single greasy pizza box in a recycling bin can contaminate the entire load. The whole batch goes to landfill, wasting every correctly sorted item alongside it.
UBC has ambitious zero-waste targets, but without reliable sorting at the source, these goals remain out of reach. The infrastructure exists. The accuracy does not.
Zero human input required.
An HC-SR04 ultrasonic sensor detects an approaching object within 30cm. The system wakes from low-power standby automatically.
An Arducam 8MP camera with a Sony IMX219 sensor captures a high-resolution image of the waste item under consistent LED lighting conditions.
A MobileNetV2 model, quantized to TFLite INT8, runs inference on a Raspberry Pi 4. Classification across six material categories completes in under 200 milliseconds.
Dual SG90 servo motors open the correct compartment: recycling or garbage. If confidence is below threshold, the item defaults to garbage to prevent contamination.
If every recycling bin on campus could sort itself, we would divert thousands of kilograms of waste from landfill each year.