Manual sorting lines in bell pepper processing are labour-intensive and expensive, and quality is difficult to manage. Through this project, Pliant will investigate the possibility to replace manned lines with smart vision and sensor technology, controlled by neural networks.
An automatic sorting system has clear benefits for growers. In addition to reducing labour costs, it will ensure the uniform quality of the peppers – which will then fetch a better market price.
Located in The Netherlands, Pliant is a specialist in high-tech 2D and 3D vision solutions.
Final report summary
Technical specifications were drawn up for a prototype using vision techniques that can generate clear and reliable pictures of bell pepper colour, wrinkles, size, shape, damage and other factors, such as the presence of aphids. This data was then used for the neural network.
The next step was to develop the prototype algorithm – the artificial intelligence to determine the quality/classification of each bell pepper based on specified terms and conditions. This self-learning software is suitable for integration into existing sorting systems.
The final project outcome was a tested and validated prototype of the technical operating principle. Results are better than expected. Individual quality observations are at least on the same level of those made by the human eye. By collecting and processing more data, the network will become more adept at recognising specific characteristics over time.