A team of Pakistani scientists has achieved a significant scientific breakthrough by creating an artificial intelligence (AI)-based visual classification method that accurately determines the sweetness of native citrus fruits. Led by Dr. Ayesha Zeb from the National Centre of Robotics and Automation at the National University of Sciences and Technology (NUST), the team successfully predicted fruit sweetness with an impressive accuracy rate exceeding 80%, all without causing any damage to the fruit.
To conduct their experiment, the researchers carefully selected 92 citrus fruits, including varieties such as Blood Red, Mosambi, and Succari, from a farm in the Chakwal district. They utilized a handheld spectrometer to obtain spectra, which are patterns from the reflection of light, from marked regions on the fruits’ skin. By employing near-infrared (NIR) spectroscopy, a technique that allows for the analysis of non-visible light spectra, the team examined the fruit samples. Out of the 92 fruits, 64 were used for calibration, and the remaining 28 were used for prediction through the spectrometer.
While the application of NIR spectroscopy in non-destructive fruit classification is not novel, the Pakistani team took a novel approach by applying it to model the sweetness of local fruits. Furthermore, they integrated artificial intelligence algorithms to directly classify the sweetness of oranges, resulting in improved accuracy.
Traditionally, assessing fruit sweetness involves chemical and sensory testing. Oranges’ sweetness is typically determined by measuring total sugars, known as Brix, while titratable acidity (TA) indicates the levels of citric acid. To develop the AI model, the team obtained reference values for Brix, TA, and fruit sweetness by extracting samples from the marked areas used for spectroscopy.
Laboratory testing of the extracted juice provided actual Brix and TA values, and human volunteers tasted the fruits and categorized them based on their perceived level of sweetness. Using the obtained spectra, reference values, and sweetness labels, the team trained the AI algorithm on a total of 128 samples. The AI model has a design to predict Brix, TA, and sweetness levels based on spectral data. To evaluate the model’s accuracy, the researchers tested it with data from 48 new fruits, comparing the predicted values with actual measurements obtained through sensory evaluations and chemical analysis.
The results were remarkable, as the AI model not only accurately predicted the values of Brix, TA, and overall sweetness but also outperformed traditional methods in sweetness prediction. The model achieved an impressive overall accuracy rate of 81.03% in identifying sweet, mixed, and acidic tastes.
This scientific breakthrough carries significant implications for the citrus industry, particularly in terms of estimating citrus fruit quality. Unlike bananas and mangoes, oranges do not continue to ripen once harvested. Therefore, this innovative AI-based method has the potential to streamline and enhance the assessment of citrus fruit sweetness, benefiting the industry and ensuring improved consumer satisfaction.
Pakistan, as the sixth-largest producer of citrus fruits globally with 0.46 million tons of exports in 2020, stands to gain immensely from this technological advancement.
The findings of this groundbreaking research have published in Nature, a prestigious research journal.
The project was a collaborative effort led by Dr. Ayesha Zeb and Dr. Mohsin Islam Tiwana from the National Centre of Robotics and Automation at the National University of Sciences and Technology (NUST), along with several other esteemed researchers from various institutions.
Moreover, Visit CxO Global FORUM or CxO News Live for all the latest updates.
