Meat quality and safety are critical concerns in the food industry, especially for products like beef and mutton, which are susceptible to spoilage and fraud. Traditional methods of assessing meat freshness and species classification, such as manual inspection, are often inefficient and prone to error. This paper introduces a novel Internet of Things (IoT) system that integrates gas sensors and advanced machine learning models, particularly deep learning, to address these issues effectively. The system combines image-based classification using a custom Convolutional Neural Network (CNN) with gas sensor data to provide a comprehensive, real-time solution for the classification of both beef and mutton in terms of species and freshness. The custom CNN, a deep learning model, was trained on a dataset of 9,928 images and achieved an impressive classification accuracy of 99%, outperforming other models such as ResNet-50, another deep learning model, and traditional machine learning models like Support Vector Machines (SVM) and K-Nearest Neighbors (KNN). The CNN showed a clear advantage in handling the complex image data, particularly in distinguishing between beef and mutton species as well as their freshness levels. The system incorporates three gas sensors—MQ135, MQ4, and MQ136—to detect gases such as ammonia (NH ₃ ), methane (CH ₄ ), and hydrogen sulfide (H ₂ S), which are released during the spoilage of meat. These gas sensor readings are utilized specifically for the classification of meat freshness. The system provides real-time feedback via LED indicators: green for fresh meat, yellow for meat that is neither fresh nor fully rotten, and red for spoiled meat. The results of the classification, based on both image data and gas readings, are displayed on an LED screen. This offers an efficient, scalable, and practical solution for real-time quality monitoring of beef and mutton. This integrated approach significantly improves the accuracy, reliability, and efficiency of meat safety management within the food supply chain.

Due to the presence of hazardous gasses at high levels, some regions are physically very difficult for humans to approach. This paper presents a project idea seeks to simplify the task of monitoring the air quality in hard-to-reach areas. This would help to know the air quality of that region. The system was developed using the Arduino Uno microcontrollers and the NodeMcu microcontroller is used in the IoT interfacing of the system. The real-time monitoring, analysis of the air quality, and data logging to a remote server, which allows for continuous data updating online, are all features of the air pollution monitoring system. The system's specified air quality readings were precise. This system detects air pollutants including Carbon Monoxide (CO), Ozone (O3), Nitrogen Dioxide (NO2), Particular Matter (PM), and Sulfur Dioxide (SO2) and the result was displayed on the LCD interface and could be accessed via the cloud on any smart device using the internet with hazardous limits indications.

Robots equipped with legs have significant potential for real-world applications. Many industries, including those concerned with instruction, aid, security, and surveillance, have shown interest in legged robots. However, these robots are typically incredibly complicated and expensive to purchase. Iron Dog Mini is a low-cost, easily replicated, and modular quadruped robot built for training, security, and surveillance. To keep the price low and its upkeep simple, we designed our quadruped robot in a modular manner. We provide a comparative study of robotic manufacturing cost between our proposed robot and previously established robots. We were able to create a compact femur and tibia structure with sufficient load-bearing capacity. To improve stability and motion efficiency, we considered the novel Watt six-bar linkage mechanism. Using the SolidWorks modeling software, we analyzed the structural integrity of the robot’s components, considering their respective material properties. Furthermore, our research involved developing URDF data for our quadruped robot based on its CAD model. Its gait trajectory is planned using a 14-point Bezier curve. We demonstrate the operation of the simulation model and briefly discuss the robot’s kinematics. Computational methods are emphasized in this research, coupled with the simulation of kinematic and dynamic performances and analytical/numerical modeling.

Social media has become a fundamental element of contemporary life, providing countless benefits but also posing substantial concerns. While technology improves connectedness and information exchange, excessive use raises issues about social and personal well-being. The emergence of social media addiction emphasizes its influence on everyday routines and mental health, with many people favoring online activities above vital tasks, resulting in real repercussions. Twitter, Facebook, and Snapchat have a significant impact on emotional well-being, adding to global rates of despair and anxiety. To measure the frequency of social media reliance, we studied data from 1,417 individuals using machine learning methods such as decision tree (DT) classifier, random forest (RF) classifier, support vector classifier (SVC), k-nearest neighbors (K-NN), and multinomial naive Bayes (NB). Understanding the behavioral patterns that drive addiction allows us to create tailored therapies to encourage healthy digital behaviors. This study highlights the critical necessity to address social media addiction as a complicated societal issue. Our major goal is to determine the amount of people who are addicted to social media.

Bananas are among the world’s most widely farmed, traded, and consumed fruits. Bananas contain several vitamins and minerals, including vitamin B6, vitamin C, magnesium, and manganese. Bananas are a very good source of vitamins and minerals in a densely populated and lower-middle-income country like Bangladesh since they are widely cultivated and available in urban and rural markets all over Bangladesh. Deep learning algorithms have shown considerable promise in recent years for image-based breed recognition tasks, encouraging researchers to look into their potential for breed-specific banana recognition. This research thoroughly investigates the use of deep learning and classical machine learning for banana breed recognition. Seven separate deep learning models were used to recognize bananas: VGG16, ResNet50, Mobile-Net, Inception-v3, MobileNetV3, Hybrid CNN-ViT, and a custom convolutional neural network (CNN), and five classical machine learning classification models: k-NN, SVM, Random Forest, Decision Tree, and Ensemble methods (SVM + Random Forest). These models were trained using images of several banana breeds to differentiate between them based on visual characteristics such as size, shape, color, texture, and skin pattern. The performance of each deep learning and classical machine learning classification model was thoroughly evaluated and tested. The customized CNN model has the highest accuracy of all models. We rigorously tested our models by comparing accuracy along with other key metrics such as recall, precision, and F₁-score. In the cases of these metrics, the customized CNN model outperforms all other deep learning and classical machine learning classification models with an accuracy of 99.37%.