| 000 | 02547nam a22002417a 4500 | ||
|---|---|---|---|
| 003 | OSt | ||
| 005 | 20261110111054.0 | ||
| 008 | 261110s2026 |||||||| |||| 00| 0 eng d | ||
| 050 |
_aT ENG _bA55 2026 |
||
| 100 | 1 | _aAninon, Ken Levy F. | |
| 245 | 1 | 0 |
_aRaspberry pi-based fruit calorie counting system using load sensor and computer vision : _ba developmental-comparative design project / _cKen Levy F. Aninon, Julius Robert R. Cortes, Juan Miguel B. Manalo, and Raha C. Pullon. |
| 264 |
_aEAC Cavite : _bSchool of Engineering, _c2026. |
||
| 300 |
_axiii, 130 pages ; _c28 cm. |
||
| 336 |
_2rdacontent _atext |
||
| 337 |
_2rdamedia _aunmediated |
||
| 338 |
_2rdacarrier _avolume |
||
| 504 | _aIncludes bibliographical references | ||
| 520 | _aCalorie tracking applications often rely on manual user input, and this can be tedious and prone to human error. Misinputs can make the whole process feel like a chore. Existing developed system also requires additional actions like shooting in different views or limitations on estimations based on height [1], [2]. To address this, this study aims to develop an automated calorie estimation system that can identify fruits and compute their calorie content without requiring text-based input from the user. The system uses a camera and the YOLOv11 algorithm to identify five common fruits, namely apples, bananas, oranges, strawberries, and kiwis. At the same time, a load cell connected to an Arduino Nano measures the fruit's weight, and calorie data is fetched from FoodData Central. Together, these components create a seamless, hands-free experience for the user. To test how well it actually works, the system was evaluated using a Developmental-Comparative research design. The fruit detection model achieved a 97% mAP, 98% precision, 94.5% recall, and a 96.2% F1 score. Calorie estimates were then compared against MyFitnessPal using paired sample t-tests. Results showed strong agreement for oranges, bananas, and strawberries, though apples and kiwis showed some differences worth noting. Overall, the system proved to be both reliable and fast, making it a genuinely useful tool for health-conscious individuals or for educational purposes. Looking ahead, the findings point toward meaningful improvements like improved sensor accuracy, a wider range of recognized fruits, and eventually, a fully offline version. | ||
| 700 | 1 |
_4author _aCortes, Julius Robert R. |
|
| 700 | 1 |
_4author _aManalo, Juan Miguel B. |
|
| 700 | 1 |
_4author _aPullon, Raha C. |
|
| 942 |
_2lcc _cTH _kEAC Cavite-Thesis _n0 |
||
| 999 |
_c100691 _d100691 |
||