Fruits provide essential nutrition in most natural form suitable for human beings. They are best when ripened naturally. However, industrialization has provided many ways for quick ripening and for extended shelf life of fruits. Detection of artificial ripening could be done by sophisticated methods like chemical analysis in lab or visual inspection by experts, which may not be feasible all the time. Of all the fruits, banana is the most consumed fruit around the world. Adulteration of banana can have devastating effects on masses on scale. It is figured, bananas are potentially ripened using carcinogens like Calcium Carbide(CaC2). In this paper, we propose and devise a novel and automatic method to classify the naturally and artificially ripened banana using spectral and RGB data. Our results show that using a Deep Learning (Neural Network) on RGB data, we achieve accuracy of up-to 90%.and using Random Forest and Multilayer Perceptron (MLP) feed forward Neural Network as classifiers on spectral data we can achieve accuracies of up-to 98.74% and 89.49% respectively.
Hyperspectral sensing has been proven to be useful to determine the quality of food in general. It has also been used to distinguish naturally and artificially ripened mangoes by analyzing the spectral signature. However the focus has been on improving the accuracy of classification after performing dimensionality reduction, optimum feature selection and using suitable learning algorithm on the complete visible and NIR spectrum range data, namely 350nm to 1050nm. In this paper we focus on, (i) the use of low wavelength resolution and low cost multispectral sensor to reliably identify artificially ripened mango by selectively using the spectral information so that classification accuracy is not hampered at the cost of low resolution spectral data and (ii) use of visible spectrum i.e. 390nm to 700 nm data to accurately discriminate artificially ripened mangoes. Our results show that on a low resolution spectral data, the use of logistic regression produces an accuracy of 98.83% and outperforms other methods like classification tree, random forest significantly. And this is achieved by analyzing only 36 spectral reflectance data points instead of the complete 216 data points available in visual and NIR range. Another interesting experimental observation is that we are able to achieve more than 98% classification accuracy by selecting only 15 irradiance values in the visible spectrum. Even the number of data needs to be collected using hyper-spectral or multi-spectral sensor can be reduced by a factor of 24 for classification with high degree of confidence
This research work was designed to evaluate the suitability and applicability of hyperspectral radiometry technology for robustly detecting adulterants in diary milk. The most common milk adulterants are (a) soda, (b) urea, (c) water and (d) detergents. The main contribution of this paper is to build a mathematical model to enable quantifying the degree of common adulterants present in milk. Data was collected using a portable spectroradiometer (Eko MS-720) which measures the spectral irradiance in the range from visible to near-infrared irradiance (350 nm 1050 nm) using samples of milk contaminated with four different adulterants (soda, urea, water and detergent) with known degree of contamination deliberately added in milk. In this study, we used the data in the range of 350 - 1050 nm to identify spectral signatures of different adulterants with different degree of concentration. Data cleansing, in the form of pre-processing was followed by machine learning techniques to create a model to capture the adulterants and also the degree of adulteration. Linear regression along with wrapper subset eval as attribute evaluator and best first search as search option was found to create the best model. Root Mean Square Error (RMSE) and Correlation Coefficient (CC) metrics were used to select the best model. The best model for detecting the degree of adulteration due to soda, urea, water and detergent in milk was found to have an RMSE of 0.027, 0.0069, 0.0382 and 0.0281 respectively while CC was 0.9919, 0.9997, 0.9887 and 0.9938 respectively. The preliminary experimental results demonstrate the effective use of spectroradiometer and machine learning technique in reliably detecting adulterants in milk.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.