Implementation of a YOLOv8 model to detect cocoa trees using Python, CustomTkinter, OpenCV and PiCamera2 on a Raspberry Pi 4 Model B.
The board used for this project was the Raspberry Pi 4 Model B (4GB) with the next OS features.
raspi@raspi:~ $ neofetch
_,met$$$$$gg. raspi@raspi
,g$$$$$$$$$$$$$$$P. -----------
,g$$P" """Y$$.". OS: Debian GNU/Linux 12 (bookworm) aarch64
,$$P' `$$$. Host: Raspberry Pi 4 Model B Rev 1.5
',$$P ,ggs. `$$b: Kernel: 6.6.62+rpt-rpi-v8
`d$$' ,$P"' . $$$ Uptime: 3 hours, 29 mins
$$P d$' , $$P Packages: 1740 (dpkg)
$$: $$. - ,d$$' Shell: bash 5.2.15
$$; Y$b._ _,d$P' Resolution: 1920x1080
Y$$. `.`"Y$$$$P"' DE: labwc:wlroots
`$$b "-.__ Theme: PiXflat [GTK3]
`Y$$ Icons: PiXflat [GTK3]
`Y$$. Terminal: lxterminal
`$$b. Terminal Font: Monospace 10
`Y$$b. CPU: (4) @ 1.800GHz
`"Y$b._ Memory: 1439MiB / 3742MiB
`"""The usage of a 64-bit OS has an important reason: it has less problems with the needed python libraries. (discovering this was really a pain :/)
The python version used for this project was: 3.11.2 (Raspberry Pi OS built-in [no need for extra installations])
raspi@raspi:~ $ python --version
Python 3.11.2In this part I am showing the used libs and its versions.
(kokoa) raspi@raspi:~ $ pip show opencv-python
Name: opencv-python
Version: 4.10.0.84
Summary: Wrapper package for OpenCV python bindings.
Home-page: https://github.com/opencv/opencv-python
Author:
Author-email:
License: Apache 2.0
Location: /home/raspi/kokoa/lib/python3.11/site-packages
Requires: numpy, numpy, numpy, numpy, numpy, numpy
Required-by: ultralytics
(kokoa) raspi@raspi:~ $ pip show ultralytics
Name: ultralytics
Version: 8.3.54
Summary: Ultralytics YOLO 🚀 for SOTA object detection, multi-object tracking, instance segmentation, pose estimation and image classification.
Home-page:
Author:
Author-email: Glenn Jocher <[email protected]>, Jing Qiu <[email protected]>
License: AGPL-3.0
Location: /home/raspi/kokoa/lib/python3.11/site-packages
Requires: matplotlib, numpy, opencv-python, pandas, pillow, psutil, py-cpuinfo, pyyaml, requests, scipy, seaborn, torch, torchvision, tqdm, ultralytics-thop
(kokoa) raspi@raspi:~ $ pip show customtkinter
Name: customtkinter
Version: 5.2.2
Summary: Create modern looking GUIs with Python
Home-page: https://customtkinter.tomschimansky.com
Author: Tom Schimansky
Author-email:
License: Creative Commons Zero v1.0 Universal
Location: /home/raspi/kokoa/lib/python3.11/site-packages
Requires: darkdetect, packaging
(kokoa) raspi@raspi:~ $ pip show picamera2
Name: picamera2
Version: 0.3.23
Summary: The libcamera-based Python interface to Raspberry Pi cameras, based on the original Picamera library
Home-page: https://github.com/RaspberryPi/picamera2
Author: Raspberry Pi & Raspberry Pi Foundation
Author-email: [email protected]
License: BSD 2-Clause License
Location: /usr/lib/python3/dist-packagesFor this project, a generic camera module was used. It has a Sony IMX219 sensor (8 megapixels)
You can buy it through the following link.
This camera must be used with the libcamera libraries, so, with that in mind, YOU MUST USE picamera2 since this lib is based on libcamera. otherwise, it wont work (even trying everything using raw opencv)
(kokoa) raspi@raspi:~/Documents/CocoaTreeDetection $ libcamera-hello --list-cameras
Available cameras
-----------------
0 : imx219 [3280x2464 10-bit RGGB] (/base/soc/i2c0mux/i2c@1/imx219@10)
Modes: 'SRGGB10_CSI2P' : 640x480 [206.65 fps - (1000, 752)/1280x960 crop]
1640x1232 [41.85 fps - (0, 0)/3280x2464 crop]
1920x1080 [47.57 fps - (680, 692)/1920x1080 crop]
3280x2464 [21.19 fps - (0, 0)/3280x2464 crop]
'SRGGB8' : 640x480 [206.65 fps - (1000, 752)/1280x960 crop]
1640x1232 [83.70 fps - (0, 0)/3280x2464 crop]
1920x1080 [47.57 fps - (680, 692)/1920x1080 crop]
3280x2464 [21.19 fps - (0, 0)/3280x2464 crop]Any uncommented situation/error: try to google it, you will almost find out everything by seeing the whole internet (this is how I solved all the errors I got) :)