School of Agriculture, Food and Ecosystem Sciences - Theses
Permanent URI for this collection
Search Results
1 - 1 of 1
-
ItemApplication of New and Emerging Technologies to Assess Rice Quality and Sensory Perception by Consumers( 2023-07)Rice is a staple food for over half of the world's population. The selections of commercial rice in the market resulted from the diversity of rice quality preferences by consumers. Besides, demographic factors such as cultural background, urbanisation, and socioeconomic status may change rice quality preferences over time. Hence, rice breeders and producers need to actively assess the best rice quality attributes that suit the demands. However, the traditional methods to evaluate rice quality are commonly tedious, time-consuming, costly and non-portable. Besides, the conventional descriptive and consumer acceptance tests based on conscious responses are likely prone to bias, expensive, time-consuming and require a dedicated laboratory sensory to conduct the sensory sessions. Therefore, new methods to assess rice quality and sensory perceptions using artificial intelligence (AI) subdivision technologies might be beneficial to overcome these disadvantages. Integrating digital sensors, computer vision, biometrics, and machine learning technologies could offer new approaches to assess rice quality and consumer perception of rice using rapid methods. This research aimed to develop rice quality assessment methods using digital sensors, computer vision, biometrics and machine learning to assess the aromas, physicochemical quality and consumer perceptions towards different types of rice. The project was divided into three main scopes to assess raw and cooked rice physicochemical quality, consumer perception of uncooked rice, cooked rice and packaging and detection of rice adulteration. This study used different types of commercial rice bought from the local markets in Australia. The rice samples consisted of regular white rice (e.g., long, medium and short-grain rice), organic, and unpolished rice from different market segments and provenances. The wide selection of rice is essential to the study since various types of rice are consumed by diverse cultural backgrounds to cook different kinds of dishes worldwide. In this study, digital, chemical and aroma fingerprints of rice were obtained using digital sensing devices such as the smartphone camera, near-infrared (NIR) spectrometer and electronic nose (e-nose). The fingerprints were used as inputs to develop machine learning models to classify different types of rice and predict the aromas, physicochemical quality, consumer perceptions, and rice adulteration levels. Furthermore, biometrics responses were used to obtain subconscious responses to assess their relationship with self-reported responses and different types of rice samples. Findings from this research are expected to contribute to the scientific knowledge of the relationship between digital, chemical and aroma fingerprinting with consumer perceptions towards different types of rice quality. Besides, the proposed method to classify different kinds of rice quality and predict rice quality attributes, consumer perceptions and adulteration levels using the integration of digital sensors, computer vision and machine learning showed high prediction accuracy (classification model: accuracy > 90%; regression model: R > 0.94). The proposed methods using the new and emerging technologies described in this study will lead the rice industry towards applying the rapid technique at a lower cost to assess rice quality compared to the traditional method.