Deep Learning-Based Method for Classification and Ripeness Assessment of Fruits and Vegetables

Applied Sciences, Journal Year: 2023, Volume and Issue: 13(22), P. 12504 - 12504

Published: Nov. 20, 2023

Food waste is a global concern and the focus of this research. Currently, no method in state art classifies multiple fruits vegetables their level ripening. The objective study to design develop an intelligent system based on deep learning techniques classify between types vegetables, also evaluate ripeness some them. consists two models using MobileNet V2 architecture. One algorithm for classification 32 classes another determination 6 overall union models, predicting first class fruit or vegetable then its ripeness. model achieved 97.86% accuracy, 98% precision, recall, F1-score, while assessment 100% 99% F1-score. According results, proposed able To achieve best performance indicators, it necessary obtain appropriate hyperparameters artificial intelligence addition having extensive database with well-defined classes.

Language: Английский

---

The use of image analysis to study the effect of moisture content on the physical properties of grains

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: May 22, 2024

Abstract Designing machines and equipment for post-harvest operations of agricultural products requires information about their physical properties. The aim the work was to evaluate possibility introducing a new approach predict moisture content in bean corn seeds based on measuring dimensions using image analysis artificial neural networks (ANN). Experimental tests were carried out at three levels wet basis seeds: 9, 13 17%. results showed direct relationship between main seeds. Based statistical seed material, it shown that characteristics examined have normal or close distribution, material used investigation is representative. Furthermore, use changes has an efficiency 82%. obtained from method this are very promising predicting content.

Language: Английский

---

Neural Network for AI-Driven Prediction of Larval Protein Yield: Establishing the Protein Conversion Index (PCI) for Sustainable Insect Farming

Sustainability, Journal Year: 2025, Volume and Issue: 17(2), P. 652 - 652

Published: Jan. 16, 2025

The predictive capabilities of artificial intelligence for predicting protein yield from larval biomass present valuable advancements sustainable insect farming, an increasingly relevant alternative source. This study develops a neural network model to predict conversion efficiency based on the nutritional composition feed. utilizes structured two-layer with four neurons in each hidden layer and one output neuron, employing logistic sigmoid functions layers linear function layer. Training is performed via Bayesian regularization backpropagation minimize mean squared error, resulting high regression coefficient (R = 0.9973) low mean-squared error (MSE 0.0072401), confirming precision estimating yields. AI-driven approach serves as robust tool yields, enhancing resource promoting sustainability insect-based production.

Language: Английский

---

Micro-Vortices and Geometric Conversion in Continuous Granular Flow for Food Powder Quality Inspection through Sensor Fusion and Computational Vision

Published: Jan. 1, 2025

Language: Английский

---

AliAmbra – Enhancing Customer Experience through the Application of Machine Learning and Deep Learning Techniques for Survey Data Assessment and Analysis

Published: Jan. 4, 2024

AliAmbra is a project developed to explore and promote high-quality catches of the Amvrakikos Gulf GP Artas’ wider regions. In addition, this aimed implement an integrated plan action, form business identity with high-added value achieve services adapted special characteristics area. The action for was actively search new markets, create collective products, their quality added value, engage in gastronomes tasting exhibitions, dissemination publicity actions, as well enhance products markets based on customer needs. primary focus publication observe analyze data retrieved from various exhibitions project, target goal improving experience product quality.

Language: Английский

---

AliAmvra—Enhancing Customer Experience through the Application of Machine Learning Techniques for Survey Data Assessment and Analysis

Information, Journal Year: 2024, Volume and Issue: 15(2), P. 83 - 83

Published: Feb. 4, 2024

AliAmvra is a project developed to explore and promote high-quality catches of the Amvrakikos Gulf (GP) Artas’ wider regions. In addition, this aimed implement an integrated plan action form business identity with high added value achieve services adapted special characteristics area. The for was actively search new markets, create collective products, their quality value, engage in gastronomes tasting exhibitions, dissemination publicity actions, as well enhance products markets based on customer needs. primary focus study observe analyze data retrieved from various exhibitions project, target goal improving experience product quality. An extensive analysis conducted by collecting through surveys that took place project. Our objective conduct two types reviews, one focused other evaluating model-driven algorithms. Each review utilized survey individual structure, each serving different purpose. our model its attention developing robust recommendation system said data. algorithms we evaluated were MLP (multi-layered perceptron), RBF (radial basis function), GenClass, NNC (neural network construction), FC (feature which used implementation system. As final verdict, determined construction) performed best, presenting lowest classification rate 24.87%, whilst algorithm worst average function). showcase expand work put into analysis.

Language: Английский

---

Efficiency of Identification of Blackcurrant Powders Using Classifier Ensembles

Foods, Journal Year: 2024, Volume and Issue: 13(5), P. 697 - 697

Published: Feb. 24, 2024

In the modern times of technological development, it is important to select adequate methods support various food and industrial problems, including innovative techniques with help artificial intelligence (AI). Effective analysis speed algorithm implementation are key points in assessing quality products. Non-invasive solutions being sought achieve high accuracy classification evaluation This paper presents machine learning architectures evaluate efficiency identifying blackcurrant powders (i.e., concentrate a density 67 °Brix color coefficient 2.352 (E520/E420) combination selected carrier) based on information encoded microscopic images acquired via scanning electron microscopy (SEM). Recognition was performed using texture feature extraction from aided by gray-level co-occurrence matrix (GLCM). It evaluated for individual single classifiers metaclassifier metrics such as accuracy, precision, recall, F1-score. The research showed that metaclassifier, well random forest (RF) classifier most effectively identified image features. indicates ensembles an alternative approach demonstrate better performance than existing traditional neural models. future, could be tool assessment products real time. Moreover, can used faster determine selection given problem.

Language: Английский

---

The improved strategy of BOA algorithm and its application in multi-threshold image segmentation

Journal of Intelligent & Fuzzy Systems, Journal Year: 2024, Volume and Issue: 46(4), P. 10471 - 10492

Published: March 8, 2024

In response to the low efficiency and poor quality of current seed optimization algorithms for multi-threshold image segmentation, this paper proposes utilization normal distribution in cluster mathematical model, Levy flight mechanism, differential evolution algorithm address deficiencies algorithm. The main innovation lies applying BBO multi threshold providing a new perspective method segmentation tasks. second significant progress is combination dynamics (DEA) improve algorithm, thereby enhancing its performance quality. Therefore, model based on optimized developed. experimental results showed that function f1, iteration improved was 53, Generational Distance value 0.0020, Inverted 0.098, Spacing 0.051. Compared with other two algorithms, has better clearer details. summary, compared existing methods, proposed effect higher efficiency, can significantly positive significance development processing technology, also provides references improvement application algorithms.

Language: Английский

---

COLOR HISTOGRAM DAN SUPPORT VECTORE MACHINE UNTUK MENGKLASIFIKASIKAN BIJI KOPI BERDASARKAN TINGKAT PEMANGGANGAN

JATI (Jurnal Mahasiswa Teknik Informatika), Journal Year: 2024, Volume and Issue: 8(1), P. 580 - 586

Published: Feb. 24, 2024

Pemilihan biji kopi berdasarkan tingkat pemanggangan menjadi faktor kunci dalam menentukan rasa yang dihasilkan. Untuk mengklasifikasikan pemanggangan, biasanya penilaian dilakukan warna dan bentuk fisik kopi. Namun, variabilitas lingkungan kondisi individu dapat memengaruhi ketepatan tersebut. Oleh karena itu, diperlukannya deteksi otomatis mengenai klasifikasi Metode computer vision dengan teknik image classification memberikan solusi terkait permasalahan Penelitian ini memperkenalkan metode processing menggunakan color histogram sebagai ekstraksi fitur untuk merepresentasikan ciri-ciri visual berbagai tingkatan pemanggangan. Selanjutnya, Support Vector Machine (SVM) digunakan algoritma mampu mengelompokkan distribusi dihasilkan dari Pendekatan bertujuan meningkatkan mengurangi pengaruh subjektif serta konsistensi proses pengklasifikasian. Hasil eksperimen menunjukkan bahwa penggabungan SVM andal akurat nilai akurasi sebesar 98,95%.

---

Usk-Coffee: A Novel Dataset and Deep Learning Benchmark Fordefect Coffee Bean Detection

Published: Jan. 1, 2024

Language: Английский

---