A
Figure 2 : Processes for deep learning in detection and maturity index classification of tomatoes in a greenhouse : ( A ) Data collection and ( B ) annotation
of correctly predicting outcomes . Thus , the ripeness index categorisation model for tomatoes was created to give consumers information on the ripeness distribution for tomatoes based on the ripeness index .
This research focused on the implementation of a deep learning technique to construct a detection and maturity index classification system for tomatoes in a greenhouse . This study involved the development of two models , with the result of the first model serving as the input for the second model . Initially , a total of 2,000 tomato picture samples were obtained through the process of image data collecting ( see Figure 2A ). Subsequently , a tomato detection model was created utilising the annotated original image samples ( see Figure 2B ). The confidence level is assessed to determine the efficacy of the primary mode , yielding a score of 95.78 %. Additionally , the tomato maturity index classification was established by utilising the original image samples that were cropped and named , based on their respective indexes . Afterwards , the accuracy of the second model was determined using a confusion matrix , resulting in an accuracy of 92.33 %. Finally , a dashboard is constructed to allow users to simultaneously monitor the whole distribution of tomatoes .
Pesticide Applicating Flying Robot ( Drone )
The increasing demand for efficient and precise agricultural practices has led to the exploration of state-of-the-art technology in the field of farming . The purpose of this preliminary inquiry
B
is to assess the feasibility and effectiveness of utilising drone ( flying robot ) sprayers for the distribution of pesticides in rice fields ( see Figure 3 ). Compared with traditional human or tractor-based spraying methods , drones are better equipped to enhance accuracy , reduce environmental impacts , and optimise resource utilisation . The research seeks to assess the effectiveness of drone sprayers in pest management in rice fields , investigating the effectiveness of using a drone sprayer to apply pesticides in a rice field in Malaysia . The study evaluated the efficacy of the drone sprayer compared to the traditional knapsack sprayer . The results indicated that there was no significant difference in the control of insect pests and diseases when comparing the use of drones with manual methods . However , there was a reduced period for spraying each plot with a drone sprayer as compared to manual application .
Figure 3 : Application “ Flying Robot ” ( Drone ) in pesticide spraying
Harvesting Robot
Before executing the harvesting task , a robot , similar to a human , must acquire the essential information regarding the nature and position of the target objects . Consequently , the advanced technology incorporated a three-dimensional ( 3D ) camera as the “ eye ” for the robot , while delicate human fingers were replaced with soft grippers . Several researchers have reported the development of a machine vision system that captures photos , identifies fruits , and determines their location . This system is designed for the goal of robotic fruit picking . These investigations encompass research conducted by [ 3 ] [ 4 ] [ 5 ] [ 6 ] [ 7 ] [ 8 ] [ 9 ].
45