AI for Strawberries: Ripeness & BRIX

Strawberries are #1 choice in summer due to distinctive sweet and sour flavour. Correctly harvesting them at the right ripeness stage and estimating their acidity and Brix values are crucial for smart farming.

AI-Driven Assessment of Strawberry Ripeness and Acidity

While many studies address ripeness classification and acidity/Brix estimation separately, the more recent joint research by Australian ???? and Korean ???? scientists demonstrates that these tasks can be effectively combined.

Using a deep learning method, the study integrates strawberry image classification with acidity and Brix estimation, achieving 96% classification accuracy and effective Brix and acidity predictions. This model outperforms other neural networks, offering significant benefits for consumers and farmers in terms of taste, harvesting, and grading of strawberries. Let's wait for implementation of this algorithm in smartphone level.

AI-powered BRIX non-destructive measurement

How sweet are your strawberries? This is the question which was asked by the Dutch ???? research team. Not just asked but answered as well in the research, published in Frontiers in Plant Science .

This research explores the prediction of strawberry sweetness, described by the total soluble solid (TSS) content or °Brix, using non-destructive and affordable hardware. The study involved collecting 13,400 images and environmental records from strawberries grown in a greenhouse over a period of several months. Various machine learning models, including convolutional neural networks (CNN), variational autoencoders (VAE), principal component analysis (PCA), kernelized ridge regression (KRR), support vector regression (SVR), and multilayer perceptron (MLP), were used to predict the Brix values based on image data, environmental records, and plant load information.

The results indicated that models trained with environmental and plant load data could reliably predict aggregated Brix values, with the lowest RMSE at 0.59. Using image data further improved the prediction of individual Brix values, achieving an RMSE as low as 1.10. The study concluded that integrating in-field images with environmental and plant-load data provides a reliable methodology for predicting strawberry sweetness, which can support decision-making in harvesting and supply-chain management. A few figures of the research are presented below.

Ripeness Assessment: which ML model is the best?

Accurate classification of strawberry ripeness stages depends on the machine learning model, which you use for this task. But.. not only! The study published by Korean ???? scientists explores the use of machine learning algorithms and various colour spaces for the accurate classification of strawberry ripeness stages. Utilizing three different machine learning algorithms, the research found that the CIELab colour space with a Feed-Forward Artificial Neural Network achieved the highest accuracy of 96.7%.

This approach demonstrated the significant impact of colour space selection on model performance and emphasized the need for further research to enhance the classification accuracy of intermediate ripeness stages.

The findings underscore the potential of integrating colour space analysis and machine learning to improve strawberry ripeness classification, thereby benefiting both consumers and producers by enabling more precise and efficient harvesting practices. Now we are expecting to see these discoveries in more smartphone applications for strawberry growers.

??What's next in strawberry tech?

Next edition of "AI for strawberries" will be focused on yield prediction for strawberries, particularly, using a combination of various data collection tools like smartphones, drones, IoT sensors, etc. Shall we focus on indoor or outdoor applications?

Please let me know in the comments ??

Wishes of sweeter and ripe strawberries,

Maryna Kuzmenko, Ph.D ???? , Chief Inspiration Officer at Petiole Pro Community

#strawberries

Photo credit for cover: Karki, S., Basak, J.K., Paudel, B. et al. Classification of strawberry ripeness stages using machine learning algorithms and colour spaces. Hortic. Environ. Biotechnol . 65, 337–354 (2024).