Design of Non-Destructive Orange Fruit (Citrus sp.) Grading System Using Near-Infrared Technology and comparison study of various regression and classification machine learning models
Skill/Tools used: python, machine learning, sklearn, tableau, matplotlib
Background
- Fruit grading is a crucial process in the food industry that involves the classification of fruits based on their quality and ripeness levels. Traditionally, this process is done manually by expert inspectors, which can be time-consuming, labor-intensive, and prone to errors.
- NIR technology works by measuring the absorption of light in the near-infrared spectrum, which can provide information about the chemical composition and physical properties of fruits. By analyzing the NIR spectra of fruits, it is possible to determine their quality and ripeness levels without damaging or altering the fruits.
- The goal of this research is to develop a reliable and efficient fruit grading system that can help improve the quality control and supply chain management in the food industry.
Data Acquisition
- A Total sample of 60 citrus fruits with different maturity are choosen and picked by local farmers, different maturity levels are based on their regular grading judgement.
- Determine the wave characteristic using 700 - 1400 nanometers wavelength resolution and determine which wavelength is most distinct between unripe fruit and ripe fruit (intensity range). I concluded that it's best to use 850nm.
- The fruits are taken to the physics lab to be measured their reflectancy intensity using 850nm NIR light.
- The fruits are taken to the life science lab to be measured using lab-tested tools
- Color: The color test on the fruit uses the Konica Minolta CR-400 Chromameter which produces an output value in the form of the CIELAB color space value.
- Texture: Fruit hardness testing was carried out using a digital Penetrometer GY-4 with a needle diameter of 3.5 mm.
- Total Soluble Solid: Testing Total Soluble Solids (TSS) on fruit was carried out using the ATAGO PAL-1 digital refractometer.
- Total Acidity: Total Titrated Acid Test (TA) was carried out by means of acid-base titration. The titration stage is carried out by taking 10 ml of slurry with a volumetric pipette, then putting it into the Erlenmeyer. 3 drops of PP indicator were dropped, then titrated using 0.1 N NaOH and observed until the solution turned purple and the amount of NaOH used was recorded.
- Organoleptic: Organoleptic testing was carried out by 30 semi-trained panelists, namely students of the Postharvest Technology Study Program. Parameters for organoleptic testing were carried out using hedonic quality parameters which assessed taste, texture, skin color, fruit color, aroma, and overall appearance.
Data Analysis and Data Science
- Comprehensive analysis are presented in this powerpoint
- Visualization using tableau for TSS Focused is here