A computer-vision system that detects and classifies fruits and vegetables in images — drawing a labelled box and confidence score around every item it finds.
This project uses a deep-learning object-detection model to locate and identify fruits and vegetables in a photo. For each item it finds, it draws a labelled bounding box with a confidence score — and it handles busy scenes with several overlapping items at once, from whole and sliced apples to broccoli, carrots, tomatoes, peppers, strawberries and kiwi.
The model is trained on the LVIS fruits-and-vegetables dataset and built with TensorFlow and OpenCV, making it a compact demonstration of a full object-detection pipeline — from a raw image to labelled, scored predictions.
Deep-learning framework powering the detection model.
Convolutional neural network for image recognition.
Image handling and drawing the detection boxes.
The fruits-and-vegetables dataset the model learned from.
binod