This study introduces an innovative method for detecting the freshness of milk through the use of smart packaging and a smartphone application. The proposed technique employs a colorimetric sensor ...array and a decision tree classifier, with a curvature test for feature selection. The freshness of milk samples was assessed using six colorimetric sensors, of which three, namely (Methyl violet, Chlorophenol red, and Nile blue) were chosen based on their classification error rates of 4%, 35.67%, and 8.33%, respectively. Sensor 1 was ultimately selected due to its superior performance, and its color information, along with the decision tree classification rules, was incorporated into a smartphone application. The application underwent an evaluation process, during which it was tested on twenty milk samples until total bacterial counts (TBC) reach 7 log CFU.g−1. The results showed that the application achieved a zero-error rate in identifying each sample, with the exception of 4–4.5 log CFU.g−1, where an error rate of 35% was observed. Overall, the total error rate for the entire evaluation period was 3%. Given these results, it can be concluded that this application has the potential to transform milk quality control and improve food safety.
•A new approach developed for detecting milk freshness using smart packaging and a smartphone application.•A colorimetric sensor array employed with Methyl violet showing superior performance.•Integrated sensor data and decision tree rules into a smartphone application.•The proposed method achieved a low error rate of 3% in identifying milk samples during evaluation.•The application has the potential to revolutionize milk quality control and enhance food safety.
•An intelligent oven for baking inspection of cupcake was designed based on image processing.•To perform cupcake classification, SVM classifier combined with rank key feature selection was ...used.•Features of G, b as well as kurtosis of a and energy were selected as optimum feature vector combinations.•The classifier accuracy obtained for G and b combination were 98.6, and for kurtosis of a and energy combination were 97.9.
In industrial ovens, the cooking process is controlled by adjusting cooking temperature and time. However, for having high-quality product, variations in parameters like inconsistency in raw materials, baking powder level, batter thickness and batter moisture should be taken into consideration. To handle these variations and produce high-quality products, an intelligent oven based on image processing was designed to scrutinize color and texture of cupcakes while baking. In this regard, a digital camera was used to take digital images from inside the oven, continuously. Subsequently, color and texture features were extracted from the set of cupcake images while baking. To maximize classification accuracy, a search algorithm along with a rank key feature using t-test class separability criteria method was implemented to find an optimum set of feature combinations. According to the feature selection algorithm, G (RGB color space) and b (Lab color space) as well as kurtosis of a (Lab color space) and energy were the most significant feature combinations. Then, the data classification was carried out using the support vector machine (SVM). The classifier accuracy, specificity, sensitivity and Kappa coefficient obtained for G and b combination were 98.6, 100, 92.3 and 95.2%, respectively; and for kurtosis of a and energy combination were 97.9, 98.3, 95.6 and 92.2% for, respectively. Pursuant to the result, an adaptive automatic control could be conducted by real-time monitoring of color and texture features of baking goods. Through this image analysis method, baking process could dynamically regulate and optimize cooking time, energy consumption as well as quality of product.
•A non-destructive vibrational method was developed to determine fruit firmness.•Vibrational features significantly correlated with fruit firmness.•A Bayesian optimization algorithm optimized the ...prediction method.•The random forest model predicted kiwifruit firmness with an RMSE of 0.0125.
Firmness has been identified as a reliable indicator for evaluating fruit quality due to its significant correlation with attributes such as taste, ripeness, shelf life and mechanical impact sensitivity. In order to prevent product loss during evaluation of fruit quality, it is essential to employ non-destructive measurement techniques. This study aimed to determine the firmness of Kiwifruit using a non-destructive vibrating method. After extracting the vibrational (amount of displacement, time delay) and physical (length, width, weight and area of a fruit) features, the most effective features were selected to predict firmness using the random forest (RF) regression method. To optimize the performance of the RF model, the Bayesian optimization algorithm was employed to identify the most appropriate hyper-parameters (number of learning cycles, learning rate, minimum number of leaves and maximum number of divisions). The results demonstrated that the developed RF model could accurately predict the kiwifruit firmness with an R2 of 0.9561 and an RMSE of 0.0125.
•All of the measurements were carried out in the laboratory as well as in the field conditions.•The measurements of investigated parameters and data transformation were conducted on-line.•Laboratory ...test measurements were performed in a real condition; therefore, the effect of tractor vibration was taken into consideration.•The main reasons of non-uniformity of seed spacing were speed of falling of seed and hitting of seed to the runner opener.•Speed of falling of seed decrease with increase of vacuum pressure value, under lower operational speeds.
Laboratory tests are usually carried out with different technologies and under quite different conditions to field tests. In this study, a laboratory test was proposed to assess seed spacing uniformity which is applicable in a field situation. In this method, a high-speed camera system was used to detect seed falling trajectory, which is an effective factor on uniformity of seed spacing in both conditions of laboratory and field. Experiments were performed with four pressures (30, 40, 50 & 60 kPa) two forward speed ranges (3 to 4.5 km/h and 6 to 8.5 km/h) and two types of seed (maize and castor) under two conditions of laboratory and field. Results revealed that the highest value of quality of feed for the seed of castor with mean of 98.31% (at forward speed of 3.0 to 4.5 km/h and 4.0 kPa vacuum pressures), and for maize seed with mean of 80% (at 3.0–4.5 km/h and 4.0 kPa) were obtained under laboratory and field conditional, respectively. To determine the locations of seed in the different frames, equations of seed trajectories were first obtained for every treatment and then a general equation was extracted from the entire data set of treatments. This equation is: y=3.523e-0.077x with value for coefficient of determination, R2 = 0.902. The results show that under operating speed range of 3 to 4.5 km/h, seed spacing uniformity for different treatments of both seeds is statistically the same at 95% confidence level both in the laboratory and in the field. At a forward speed of 6 to 8.5 km/h and 4 levels of vacuum pressure, the difference among calculated mean and actual mean of speeds of fall is statistically significant at 5% probability level.
•Sound analysis was used to detect the carob moths in pomegranate fruit.•Different feature selection methods along with Fischer classifier were applied.•Features of natural frequency, spectral ...entropy and cross-over were selected.•The accuracy of the classification was 97.55%.
Sound analysis is one of the least destructive testing methods that have been developed for the determination of fruit quality. Therefore, in this research, a system was developed for analyzing the acoustic impulse response to determine the presence of carob moths in pomegranate fruit. For this purpose, acoustic signals generated from 100 pomegranate fruits were recorded (50 healthy and 50 infected). In order to record and extract the acoustic signals' characteristics (non-destructive test), a plastic pendulum hit the samples at three speeds (0.3, 0.9, and 1.5 m/s) and a microphone was located at three positions concerning impactor (0, 90, and 180°). For data classification, Fisher's linear discriminant classifier combined with three different feature selection methods was used. According to the statistical analysis, a speed of 0.9 m/s was selected as the optimum impact velocity. According to the classification results, the best features were natural frequency, spectral entropy, and zero-crossing, which had the least classification error. The application of these three features resulted in a classification accuracy of 97.55%.
•A computer vision system was designed and developed to evaluate the browning process of banana peel.•Method of digital image processing was effective to prediction and investigation of POD and PPO ...enzymes during browning.•Two equations were obtained using GP modeling to predict and detect the changes of the activity of the PPO and POD enzymes.•The accuracy of proposed method confirmed through comparisons with enzyme activity (PPO and POD).•The color features which were extracted from images of samples, play an important role in evaluation of browning process.
The process of enzymatic browning is one of the most important chemical reactions, which effects on color, appearance and quality of fruits and vegetables. Polyphenol oxidase (PPO) and peroxidase (POD) enzymes are associated with enzymatic browning in the tissue of agricultural products. Quality of banana as a climacteric fruit is reduced by enzymatic browning during storage. Therefore, to evaluate enzymatic browning in banana, first, images of the fruits were taken at 25 °C for 9 days. Then, these images were investigated using digital image processing in order to predict and study POD and PPO enzymes during the browning process of banana peel. To this end, seventeen color parameters (R¯, G¯, B¯, VR, VG; VB, r, g, b, C1, C2, C3, C4, C5, C6, C7, C8) were extracted from each image as non-destructive parameters. In the following, PPO and POD both were measured through the laboratory methods Finally, using genetic programming (GP) modeling, two equations were obtained which can be used to predict and detect the changes of the activity of the PPO and POD enzymes during the storage period (9 days). The correlation coefficients between the measured values and the predicted values for PPO and POD enzymes were 0.98 and 0.97, respectively. Furthermore, there were no significant differences between predicted values with measured values of PPO and POD enzymes (p > 0.05); these results indicate the proper performance of the designed models.
•An automatic control system has been designed to monitor nutrient levels in plants.•Software was developed to detect and control iron deficiency in spinach plant by extracting color, morphological ...and texture features from images.•The system can determine the plant’s iron needs by analyzing six extracted image features.•A relationship has been found between the extracted image features of the plant’s iron content.•One factors that affects the concentration of chlorophyll in spinach is its iron content.•Intelligent control can reduce iron nutrient solution consumption.
The regulation of nutrients intake in plant life is a critical factor in preserving the quality of crops, enhancing yield rates, and minimizing the fertilizer consumption. Consequently, this research presents the design of an intelligent robot for real-time monitoring of plant nutrients within greenhouse environments. The robot’s ability to detect and address iron deficiency in spinach plants was evaluated. To achieve this, the robot, equipped with a camera, captured visible imaging from surface of spinach plants including both the control group and the group subjected to iron deficiency stress. Statistical analysis of the data was performed at the probability level of 5 %, and genetic algorithm (GA) was utilized to select the most suitable features (G, Area, Energy, Entropy, R and L), from 41 extracted image parameters, including color, morphological, and texture features. An artificial neural network (ANN) was then employed to detect iron deficiency in plants, resulting in an accuracy of 96 %. Subsequently, the Gaussian kernel function of support vector regression (SVR) algorithm was used to predict spinach iron content, yielding the highest determination coefficient (R2 = 0.94) and the lowest values of mean absolute percentage error (MAPE = 5.92), root mean square error (RMSE = 0.43) and standard regression error (SRE = 2.66). The designed robot detected iron-deficient plants with precision, sensitivity, specificity and accuracy of 86 %, 82 %, 84 % and 83 %, respectively. The robot then took the necessary control actions to supply the deficiency of this nutrient by foliar spraying of more than 89 % of the crop area.
The majority of the volume in a plant cell is water. Therefore, the growth and metabolism of plants are highly dependent on the changes in plant water content. To optimize plant growth in water ...limited and drought stress conditions, many mechanisms have been considered. The aim of this study was to design and develop an intelligent system, based on an Artificial Neural Network (ANN) and machine vision that would optimize plant growth in limited water situations. To this end, color, morphological and textural features were extracted from a set of turfgrass plant images under drought stress conditions and were analyzed to determine plant water requirement. To maximize classification accuracy, an optimum set of features h (hsl color space), L (Lab color space), H (HSV color space) and PDF1 (Probability Density Functions) were selected using a genetic algorithm. Then a data classification operation was conducted using an ANN. The classifier accuracy for three plant situations (fresh, at the edge of wilting and wilted) as well as its total accuracy were 91.3, 77.8, 97.9 and 90.7%, respectively. The automated irrigation system could measure and determine the plant wilting condition by investigation of four extracted features and then determine and apply the correct amount of water required for optimum plant growth in water limited situations.
The aim of this study is the detection of adulteration in honey by a microcontroller measurement device. For this purpose, 18 pure honey samples from Sidr, Locoweed and Citrus honey were prepared, ...whose physicochemical characteristics, including moisture content and ⁰Brix, pH, color and ash content were measured. To detect impurities in honey, electrical conductivity and light refraction were used as inputs of an artificial neural network (ANN) classifier. This allowed fraud detection in honey by a portable instrument developed by enclosing the electrical conductivity and light sensors connected to a smartphone application. The results revealed that increasing impurities caused a decrease in ⁰Brix and an increase in the pH of honey. The highest value of moisture was 17.5% (for Locoweed) and the highest value of pH was 4.69 (for Sidr) at 75% impurity concentration. Citrus showed the highest value of ash and electrical conductivity of 0.36 and 415 ppm, respectively. The lowest error of training the ANN classifier of 13% was obtained with 14 neurons as the optimal number of neurons in the hidden layer. The honey samples with fructose concentrations of 0% (pure honey), 10%, 25%, 50%, 75% and 100% (pure fructose) were successfully classified, with correct classification rates of 100%, 83.3%, 80%, 87.5%, 100% and 100%, respectively. This reveals the acceptable performance of the designed device in detecting honey adulteration.
•A simple and low manufacturing cost system was proposed for honey fraud detection.•High and reliable classification accuracy (91.45%) was achieved using ANN classifier.•Transmitted light and electrical conductivity correlate with %fructose in pure honey.•Android application developed to collect data from sensors and show real time purity.
Eggs are a good source of high quality protein and knowing their quality (physical and chemical properties) during storage is of great importance. Thus, the aim of this research was to design a ...computer vision system to assess egg freshness during storage time. To this end, 210 intact eggs were collected and stored for 30 days under room conditions (25 ± 2 °C and 20 ± 3%). After imaging, every other day, some internal and external quality characteristics including yolk height, yolk and albumen pH, yolk and albumen density and Haugh unit (HU) were measured as destructive parameters and area index (D) egg weight as non-destructive parameters. Based on Pearson correlation coefficients, area index were significantly correlated with all destructive variables (
p
< 0.05). In order to predict egg freshness, artificial neural network was trained by Levenberg–Marquardt, scaled conjugate gradient, Bayesian regulation, resilient and radial basis algorithms. The best result of artificial neural network for HU and albumen pH prediction was achieved by the Levenberg–Marquardt algorithm with the correlation coefficient of 0.93 and 0.87, respectively.
