Advanced driver-assistance system on Jetson Nano with 3 neural networks

[vietanhdev]

(Demonstration video is in the first post below because I only can add 3 links per post).

Currently, I built an advanced driver-assistance system on Jetson Nano with four main functions: forward collision warning, lane departure warning, traffic sign recognition and overspeed warning. In this project, I trained and optimized three deep neural networks to run simultaneously on Jetson Nano (CenterNet-ResNet18 for object detection, U-Net for lane line segmentation and ResNet-18 for traffic sign classification). Checkout my project and the source code in following blog posts.

• Introduction and hardware design: https://aicurious.io/posts/adas-jetson-nano-intro-and-hardware/
• Software design: https://aicurious.io/posts/adas-jetson-nano-software/
• Deep neural networks: https://aicurious.io/posts/adas-jetson-nano-deep-neural-networks/

Look forward to seeing your comments!
Best,
vietanhdev

[therealleonmusk]

Nice, Some one else has a project for driver situational awareness floating around here too. Couple of nano’s and you’ve got a nice lil interior and exterior management system. :)

[dusty_nv]

Cool project, thanks for posting! Nice work with all your DNNs and the performance on Nano.

[jaybdub]

Awesome work @vietanhdev!

Great blog post on CenterNet-Resnet18 as well.

Best,
John

[MadRobotics101]

Is this project open source? Also, what type of algorithm are you using (e.g., CARS, ACARS, etc)?

[marcomag]

This looks like a nice project, However, I cannot use a mouse to control the GUI, nor a web camera as a video input device, it looks like it can be used only with the video examples provided. I’m having to order a Waveshare touch screen to control the project. I tried finding the camera configuration but I haven’t been able yet.

Can you provide some instructions to set up a web camera?