How to configure your Raspberry PI 4 to complete autonomous driving object detection on the highway!
Exclusive thanks to EdjeElectronics and with his tutorials, for the inspiration and greater part of the code for this job:
https://github.com/EdjeElectronics/TensorFlow-Lite-Object-Detection-on-Android-and-Raspberry-Pi/blob/master/Raspberry_Pi_Guidebook.md
Commitment
========
The target of this task was to teach a neural community to detect items on the street that an autonomous driving car or truck would see (eg. bus, visitors light-weight, site visitors sign, particular person, bike, truck, motor, automobile, practice, rider). Then to examination the educated community on lightweight components (i.e. Raspberry PI 4) to see how it performs in conditions of processing pace and detection accuracy.
Far more Aspects on Community Coaching
===========================
https://ecd1012.medium.com/autonomous-driving-object-detection-on-the-raspberry-pi-4-175bba51d5b4
Specifics of Application and Neural Community Model for Object Detection:
=======================================================
– Language: Python
– Framework: TensorFlow Lite
– Network: SSD MobileNet-V2
– Coaching Dataset:Berkely Deep Push (BBD100K)
Instructions:
==========
sudo apt-get update && sudo apt-get up grade
git clone https://github.com/ecd1012/rpi_highway_item_detection.git
cd rpi_street_item_detection
sudo pip3 put in virtualenv
python3.7 -m venv TFLite-venv
source TFLite-venv/bin/activate
bash get_py_specifications.sh
sudo raspi-config
Working Detection:
===============
python TFLite_detection_webcam_loop.py –modeldir=TFLite_model_bbd –output_path=processed_illustrations or photos
More Data:
==================
* This url explains what virtual environments in python are and why it’s a excellent thought to use them when acquiring: https://towardsdatascience.com/digital-environments-104c62d48c54
