Mushkin: Enhancing mushroom yield production via long short-term memory (lstm) with monitoring using yolo v8 for image processing in an iot-based urban agriculture/ Robin John T. Cabrillas, Karl C. Cabulagan, Adrian C. Dela Cruz, Ayessa Denise S. Opanda, John Harold S. Ricafrente, and Maica V. Tameta.--

Bachelor's thesis

College of Engineering.-- Bachelor of science in electronics engineering: Technological University of the Philippines, 2025.

Includes bibliographic references and index.

Millions of people in the Philippines continue to struggle with food insecurity. Growing
mushrooms has become a popular way to increase food production in constrained areas
through urban agriculture. This paper presents MushKin, an Internet of Things (IoT)-based

system that combines YOLO v8 with Long Short-Term Memory (LSTM) models for real-
time monitoring and yield optimization in mushroom farming.

Environmental sensors are used by the system to track variables like light intensity,
substrate wetness, CO2 levels, temperature, and humidity. While LSTM algorithms
estimated ideal growing conditions, the ESP32 microcontroller gathers data to enable
autonomous climate management. YOLO v8 also makes it easier to identify issues and
recognize growth stages, which improves cultivation efficiency. To provide accessibility
and user-friendliness, a web-based platform enables farmers to remotely monitor and
control the system.
MushKin was tested and evaluated at Bigay Buhay Multipurpose Cooperative (BBMC)
to compare it with conventional farming practices. Results show that MushKin
significantly increases sustainability, decreases labor-intensive monitoring, and improves
yield prediction accuracy for five different mushroom species: oyster, milky, reishi,
chestnut, and shiitake. The MushKin method has been shown to enhance productivity,
according to statistical research, including T-tests.
The precision and productivity of the Mushkin system were shown to be significantly
higher than those of conventional mushroom cultivating techniques. Reliable real-time
detection and measurement were made possible by the model's 0.707 precision, 0.637
recall, and 0.673 [email protected], all of which were obtained using YOLOv8. When compared

to hand measurements, size estimations of cap area, length, and diameter for five different
species of mushrooms consistently had percentage errors below 7%, demonstrating the
consistency and quality of the approach.
Additionally, significant increases were found when comparing the yields of Mushkin
and conventional farming methods. Cap size, stem length, and total harvested weight all
showed statistically significant changes, favoring tradition supported by Mushkin.
Extended harvesting times were also made possible by Mushkin-enabled systems, which
raised the overall yield. T-tests confirmed these results, confirming the system's ability to
improve quality and productivity.
This study demonstrates how AI-powered IoT solutions can be used in urban agriculture
to support sustainable food production. MushKin offers an efficient and effective method
of mushroom farming by combining automation, predictive analytics, and smart
monitoring, opening the door for more extensive precision agriculture applications.