| Formatted contents note |
ABSTRACT: STATEMENT OF THE PROBLEM: As the pandemic hit the world, many families have decided to include a tiny garden from within their homes due to different personal reasons such as saving bills to produce their own vegetables, or simply just a hobby to get through their day. According to Yara International (2022), the world’s population is increasing, and we will number 9 billion by 2050. By then, we’ll need to produce 60% more food on the same amount of land. Achieving flood security necessities having enough healthy flood available at reasonable rates at all times, regardless of where people reside. Therefore, the researchers aim to develop a system and device that can accurately guide home gardeners to growing their tomato plants healthily using the Optimal Plant Fertilizer Application Featuring an Automatic Dispenser Throughout a Tomato Plant’s Growth Stage using Artificial Network. The study will answer the following questions: 1. Can the researchers successfully build a hardware device that can efficiently and automatically dispense fertilizer as the system commands? 2. How will the researchers construct the software system using the Phyton programming language that will be the base operation for the entire system that shall achieve their desired goals and objectives? 3. Can the system properly work with precision and efficiency in mind which can significantly contribute to the daily living of the home gardeners, most especially in growing their tomato plants? RESEARCH METHODOLOGY: The research to test the initial theory and determine if it is true or false, and the study’s objective and structure are predetermined from the start. As a result, when the suggested device is evaluated, its proponents will obtain feedback using a data collection approach, analyze the resulting information, and determine whether the study’s purpose was reached. The researchers will move forward with the experimental research strategy for this study. The research will serve as a test to properly document and analyze the working parameters needed in determining the growth stages of a tomato plant. It will use the different hardware components and software programs to analyze the different tomato parameters and conclude from that about the tomato plant’s current growth stage. There, the appropriate fertilizer for that growth stage will then auto-dispense in set-time. SUMMARY OF FINDINGS: This research aims to create an automated fertilizer dispensing system where it will be able to detect a tomato plant’s current growth stage and supply the appropriate fertilizer requirements without any human intervention. It also specifically aims to answer the following: 1. To develop s system that can automatically determine the current growth stage of a tomato plant using deep neural network. • To create models that can properly identify the different growth parameters of a tomato plant. 2. To construct the hardware that can automatically dispense fertilizer. 3. To test functionality of the system and discover the difference between system-assisted fertilizer application versus manual application. Based on the following research problems, the researchers found the following results from the experimentation as follows: 1. With the help of Roboflow’s annotation tools and machine learning model training, the researchers have successfully created the following tomato plant parameters detection model: Stem Model, Leaf Model, and Fruit Model. The gathered accuracy or precision rate for the Stem Model is at 75%, while the Leaf Model has a rate of 88.3% and lastly, for the Fruit Model, the rate gathered is at 80.7%. This is against the data included in the datasets. 2. Next, the researchers have successfully integrated the software program flow to the constructed hardware for the automation of not only the fertilization for each growth stage, but also, as an additional feature, the automation of watering every 8:00 AM. Everytime the camera captures a photo at 8 AM, it will send the data to the database in which the program will then analyze using trained data modeled from Roboflow, after it will then convert it to centimeters which will be used to determine the current growth stage. Lastly, it will then reroute back to the hardware in which it will open the appropriate servo motor to let in the proper fertilizer to flow inti the plants. 3. Lastly, the researchers then tested the difference between machine-assisted fertilization against normally-fertilized, and unfertilized sets of plants. a. The researchers calculated that the set with Automatic Fertilization for the Stem Parameter has a growth percentage of 14.58% and is the highest, compared to the other sets, while for the Leaf Parameter, the Manual Fertilization is the highest at 17.58% compared to the Automatic which is at 17.49%. b. While for the accuracy of the models in contrast to manual measurement, the researchers found that for the stem and leaf percentage error rate, the calculated value is at 18.21% and 13.07% respectively. Meaning that for an 8 cm stem, there is a deviation of +- 1.45cm, while for a 4 cm leaf, there is a deviation of +-0.52cm compared to the manual measurement. CONCLUSION: Tomato planting can be challenging specially for beginners and for those who want to further their knowledge in growing these plants. The next step is the proper application of fertilizer, especially the correct type of it for every growth stage. This research then concludes that having the knowledge on the application of research is beneficial to the growth rate of a plant and not only specific to tomatoes. According to the gathered data and findings, automated fertilizer can help boost the growth rate of a tomato, not only that, since automation is at work here, this can be further used in large-scale tomato farming that can help further the research in growing tomatoes. The results gathered are also highly evident of a positive difference between machine-assisted versus manual application of fertilizer although the difference is small, but nonetheless, it is still positive. RECOMMENDATION: In liu of the conclusion of this research, the researchers present the following recommendations for further researching: • Conduct further research regarding the trained models since the precision rate of some components is lower than 80. The researchers suggest researching more about the trained models and increasing the images within the datasets to further increase the accuracy rate for each model. • Test the research with more knowledge regarding tomato planting and to also test the trained model against a tomato plant from the germination stage up to the blooming stage. • Further refine the program flow of the research and also include other possibilities of hardware set-up where the trained model can not only detect front-row tomato plants, but also a huge array of plants. • Lastly, develop the research further by conducting additional studies and gathering more data to determine new functionalities and features such as inclusion of pest-detection and disease-detection to incorporate the automation of fertilization with other aspects of planting. |