I'm working on same project, the technique I'm using: Segmentation model yolov8 train on lanes Detection model to detect vehicles Get mask and bounding boxes then check the intersection of lane mask with bounding boxes to get number of vehicles in each lane.
SSD model by default can detect all the objects in the coco dataset so incase you have different objects rather than the ones in the coco dataset then you have to fine tune your custom model using SSD pretrained model
Lets suppose we have a one frame , which is image of the scene. so how can we count how many vehicles in a lanes , in video its quite simple we can track but for the image what will be the solution you think ?
@@tech_watt how this approach works on every image ? Ai is to reduce the manual tasks, we cannot define zones for every image ,if the camera is moving then what we should do ? Let's work on a general approach
![เปิด อาณาจักร หมื่นล้าน "สาระตั้ม" l [Nickynachat]](http://i.ytimg.com/vi/3XE0nPF4YkQ/mqdefault.jpg)
I'm working on same project, the technique I'm using:
Segmentation model yolov8 train on lanes
Detection model to detect vehicles
Get mask and bounding boxes then check the intersection of lane mask with bounding boxes to get number of vehicles in each lane.
I have not tried on video right now .
That's fantastic am grateful my tutorial helps.
@@tech_watt no i have seen your video now . i will try your technique , lets see how much accuracy we get on a static camera
Amazing Explanation . Love from India
Thanks for watching
amazing! can we use 2 live cctv cameras an input? we are making a smart traffic light system
Yes that’s very possible
@@tech_watthow can we get the results and send it to an arduino code? can you help me?
Awesome 🎉
Hi there! How would I change the code to have SSD detect various things on a video, and not on an image?
SSD model by default can detect all the objects in the coco dataset so incase you have different objects rather than the ones in the coco dataset then you have to fine tune your custom model using SSD pretrained model
Share the video link used in the project
video link is in the video description
Video Link: www.pexels.com/video/traffic-at-brooklyn-queens-expressway-5229647/
This is great
Thanks for your feedback i really appreciate it
how you install lap ? bcz it gave me error when i try to install it'
pypi.org/project/lap/
Lets suppose we have a one frame , which is image of the scene. so how can we count how many vehicles in a lanes , in video its quite simple we can track but for the image what will be the solution you think ?
at 0;02 we have 5 vehicles in Lane A, 8 in lane B, 9 in lane C
0:02
With image we can still use Same approach but then we just have to tweak the code in other to count all the ids in a particular zone
@@tech_watt how this approach works on every image ? Ai is to reduce the manual tasks, we cannot define zones for every image ,if the camera is moving then what we should do ? Let's work on a general approach