Joseph (O) Fatoye

Joseph (O) Fatoye

As an aspiring Mechatronics Engineering professional, I am passionate about pushing the boundaries of the field of Cognitive Robotics.

Object Detection Fruit Sorting with Edge Impulse

less than 1 minute read

Published:

Description

Trained a neural model to identify and classify simple fruits in the immediate environment. 90% training performance and 84.62% real-world success rate. This was a class Project.

Task

“Create a model that identifies and correctly classifies objects, and in real time. For this course we’ll be grouping you all in groups of tens
…We’ll be using Edge Impulse for this project as well.”
– Dr. Segun Adebayo

My group was tasked with sorting oranges and cherries.

Steps

  • Gathered as a group
  • Assigned individual tasks
  • Bought a lot of oranges and cherries
  • Started training the neural model

Challenges Faced

  • Off season for oranges

    By the way, are oranges called oranges because oranges are orange?👀

  • Variable lighting conditions in training image data
  • Classification problems for similar looking fruits

Results

  • Last training performance (validation set) F1 SCORE = 90.0%
  • Model testing results: ACCURACY = 84.62%
  • Lots of oranges and cherries to go around for group members 😁

Training Performance Results Training Performance

Real-world Data Results Real-world Model Testing Results

Acknowledgements

  • Dr. Segun Adebayo
  • My group members

You May Also Enjoy

Tele-operated Surveillance Bot

4 minute read

Published:

Description

I collaborated with AFEME students to assemble a tele-operated bot using ESP32 CAM, Arduino Uno, L298n motor driver, Lipo batteries, and DC motors. Programmed the bot’s functionality in C++ via Arduino IDE, with Firebase for real-time communication and React.js for a user-friendly software interface.

Sunflower Solar Tracker System

2 minute read

Published:

Description

I captained a team of four Mechatronics Engineering undergraduate colleagues to build a single-axis solar tracker system that maximizes solar energy generation, leveraging Arduino and accessible electronic components. This was a class project.