On this challenge we see tips on how to construct a tool that detects maturation levels based mostly on colour with a neural community mannequin. As vegatables and fruits ripen, they alter colour as a result of 4 households of pigments: chlorophyll (inexperienced), carotenoids (yellow, crimson, orange), flavonoids (crimson, blue, purple), betalain (crimson, yellow, purple).
These pigments are teams of molecular constructions that soak up a selected set of wavelengths and replicate the remainder. Unripe fruits are inexperienced as a result of chlorophyll of their cells. As they mature, the chlorophyll breaks down and is changed by orange carotenoids and crimson anthocyanins. These compounds are antioxidants that forestall the fruit from spoiling too rapidly within the air.
After performing some analysis on colour change processes throughout fruit and vegetable ripening, we determined to construct a synthetic neural community (ANN) based mostly on the classification mannequin to interpret the colour of fruit and greens and predict ripening levels.
Earlier than constructing and testing the neural community mannequin, we developed an online software in PHP (operating on a Raspberry Pi 3B+) to gather the colour information generated by the AS7341 seen gentle sensor and create a dataset on the maturation levels . We used an Arduino Nano 33 IoT to ship the produced information to the online software.
After finishing the dataset, we constructed the unreal neural community (ANN) with TensorFlow.