Flucus / Traffic-Sign-Detection-AI

AI-driven traffic sign detection and recognition system using YOLOv8 and a diverse dataset including Hong Kong traffic signs, designed for enhancing road safety and efficient traffic management.
Apache License 2.0
1 stars 1 forks source link

Can't access the training dataset via the link #2

Open Flomedi opened 5 months ago

Flomedi commented 5 months ago

Hello there,

I used your trained model in a project of mine and must say I got very good results right from the start! Thanks for sharing the model :)

I would like to further fine-tune the model in order to detect my local signs (Europe/Germany) with better accuracy - speed signs seem to be a recognised a little bit inconsistently.

The goal of my project is to fit a speed limit recognition system and possibly more to an old car (Mercedes W201).

Here is a screenshot from a video, where I ran live-inference using my phone as a webcam in the car, securely mounted using duct-tape on the rear-view mirror :D

vlcsnap-2024-04-22-20h49m32s071

I would be very happy, if you could provide me with the necessary config file, as well as the folder with training/validation images/annotations!

Greetings from Germany, Flomedi

sumitshrestha commented 3 months ago

Yeah i would love to do similar work for US based signs. Any instruction on fine tuning would be appreciated

Flomedi commented 3 months ago

Yeah i would love to do similar work for US based signs. Any instruction on fine tuning would be appreciated

I the meantime, I've gathered some other datasets from the web and am planning on training models on those.

I'm also thinking of creating an open source project regarding my project - im a beginner in this field and I think that help from outside could make my dream of my old car being Tesla like come true :)

Would you be interested in collaborating in an open source project of this kind?