Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Not able to count the vechile #35

Open
venkateshbabusekar opened this issue Feb 8, 2019 · 31 comments
Open

Not able to count the vechile #35

venkateshbabusekar opened this issue Feb 8, 2019 · 31 comments

Comments

@venkateshbabusekar
Copy link

I am trying to use my own video which is 30 frames per second and resolution is of 1920*1080. I tried to change the ROI value from 200 to different values but it didnt work.

Can you please guide me on the same??

I would also like to know where to update the frames per second value.

Looking forward to hearing from you.
@tech123master
Copy link

Hi. I am having the same issue. When I use a different video I am not able to get any output. I tried to change ROI values but even then it did not work.

Would be nice to get a guide on how to run our own traffic videos using this project. Thanks :)

@harshsp31
Copy link

@venkateshbabusekar @ahmetozlu Did you get any solutions for this?

@venkateshbabusekar
Copy link
Author

@harshsp31 Nope. I didn't have time to look into the code. Please do let me know if you're able to find the solution for the problem.

@ChampionTej05
Copy link

There are 3 files, which are concerned with change in ROI. Firstly, since the code provided in vehicle_detection.py is done hardcoded for sample video, as written in comments. So to make it work for your own video,

  1. find the below in vehicle_detection.py file. Here in putline function, instead of (640,200) write (widthOfFrame, ROI_Location).
if counter==1:
cv2.putline(...)
  1. The second change needed is in utils.visualization file, change the value of ROI to your suitable values
  2. Somehow, utils.speed_prediction.predict_speed function is not been able to get the ROI value passed to it by utils.visualization file, hence just manually set roi_position=200 in the function.
    With above 3 steps you will be able to do it for your own video.

In the sample video, Direction of vehicles is DOWN, so if your video has UP direction (Back Face of Vehicle) then you have to change the logic in utils.speed_prediction.predict_speed function. Find the line
where is_vehicle_detected.insert(0,1) is written. In the IF condition, instead of (bottom) just write (top) and also 10 lines down, where speed is calculated, change bottom_position_of_detected_vehicle.insert(0,top)

If you need the modified code, let me know.

@venkateshbabusekar
Copy link
Author

Hi @ChampionTej05 It would be great if you could share the modified code.

Looking forward to hearing from you.
Best,
Venkatesh

@ChampionTej05
Copy link

ChampionTej05 commented Jun 18, 2019 •
edited

@venkateshbabuekar this is the link for the zip file of the folder. Make sure you place the files in this folder in the research directory of object api. The accuracy of counting solely depends upon whether your model is able to detect the vehicle near the ROI. So analyze your video once and see which line in the frame is best suited where you get maximum detection.

https://drive.google.com/a/rknec.edu/file/d/1-6KgDYd1HPOdJc4okgcfDgpscbhNfItP/view?usp=drivesdk

@amaadmirza
Copy link

@ChampionTej05 give me access of your google drive to download code ....

@ChampionTej05
Copy link

ChampionTej05 commented Jun 22, 2019 via email

@amaadmirza
Copy link

@ChampionTej05 i want to change the ROI line vertically to detect vehicles and counting not horizontally kindly suggest possible settings .... waiting for your reply

@amaadmirza
Copy link

@ChampionTej05 waiting for your reply....

@ChampionTej05
Copy link

ChampionTej05 commented Jun 24, 2019 via email •
edited

@amaadmirza
Copy link

@ChampionTej05 thanks for your help .... kindly share your sample video through google drive

@ChampionTej05
Copy link

Video on which I am running my project is confidential, so I can't share it. You can use any video of traffic.

@amaadmirza
Copy link

ok no problem.. any other link of other traffic video that related to sample code .....!!

@ChampionTej05
Copy link

Leave your email, I will share you personally.

@amaadmirza
Copy link

my email id: amaadmirza@gmail.com

@amaadmirza
Copy link

@ChampionTej05 where we can add kalman filter in this code ? i need to add kalman filter...

@ChampionTej05
Copy link

Provided you know how to add the kalman filter on Open CV Frame, you can just go and check out the line 200 in roi_detection.py file. input_frame variable is your Frame. Just add filter to this frame.

@amaadmirza
Copy link

my question is about how to use because the object is bumping in the frame... is it apply to the frame or bounding boxes like this

            newboxes = []
            for box in boxes:
                box=kf.KalmanFilter.predict(box)
                kf.KalmanFilter.update(box)
                newboxes.append(box)
            boxes = newboxes

@ChampionTej05
Copy link

According to the Documentation it should be applied to the boxes.

@amaadmirza
Copy link

amaadmirza commented Jun 26, 2019 •
edited

@ChampionTej05 i am using this github code

https://github.com/zziz/kalman-filter

for kalman filter but not successful could you plz help me to put kalman filter in your code...

@amaadmirza
Copy link

@ChampionTej05 tried this kalman filter but not successfully integrate in it ..... need your help

@i-am-manish
Copy link

i-am-manish commented Jun 29, 2019 •
edited

This are the listed changes you would need. Roi_detection.py : ROI_POSITION= 1300 (you can set according to your frame size, if your frame size 1920 then set it to 1300) Around 277 line : cv2.line(input_frame, ( ROI_POSITION,0), (ROI_POSITION,height_img), (0, 0, 0xFF), 5) Visualization_utils.py: ROI_POSITION: should be same as above Around 181: Change the IF condition to : left < ROI_POSITION Speed_Prediction.py: Roi_position: Same as above Changes around line 39: many changes hence I am attaching file for it. I have attached other files also. You can check. Vehicle_tensorflow

This are the listed changes you would need. Roi_detection.py : ROI_POSITION= 1300 (you can set according to your frame size, if your frame size 1920 then set it to 1300) Around 277 line : cv2.line(input_frame, ( ROI_POSITION,0), (ROI_POSITION,height_img), (0, 0, 0xFF), 5) Visualization_utils.py: ROI_POSITION: should be same as above Around 181: Change the IF condition to : left < ROI_POSITION Speed_Prediction.py: Roi_position: Same as above Changes around line 39: many changes hence I am attaching file for it. I have attached other files also. You can check. Vehicle_tensorflow

Sir, can u provide access to this drive link
mailto : hey_manishp@gmail.com
thanks

@ChampionTej05
Copy link

I have made the link public

@i-am-manish
Copy link

i-am-manish commented Jun 29, 2019 via email

@sidharthskumar
Copy link

I have made the link public

Sir, couldn't access says it requires permission

@haiqalma
Copy link

haiqalma commented Aug 1, 2019

@ChampionTej05 i have changed all the roi position in vehicle_detection_main.py, speed_prediction.py and visualization_utils.py but still cannot counting vehicles, help me please :(
error

@linchunmian
Copy link

hi, @ChampionTej05 please give me access of your google drive to download code. Thanks in advanced.

@ZIKO94ZIKO
Copy link

@ChampionTej05 could you explain this lines please :

if isInROI: pixel_length = bottom - bottom_position_of_detected_vehicle[0] scale_real_length = pixel_length * 44 # multiplied by 44 to convert pixel length to real length in meters (chenge 44 to get length in meters for your case) total_time_passed = current_frame_number - current_frame_number_list[0] scale_real_time_passed = total_time_passed * 24 # get the elapsed total time for a vehicle to pass through ROI area (24 = fps) if scale_real_time_passed != 0: speed = scale_real_length / scale_real_time_passed / scale_constant # performing manual scaling because we have not performed camera calibration speed = speed / 6 * 40 # use reference constant to get vehicle speed prediction in kilometer unit current_frame_number_list.insert(0, current_frame_number) bottom_position_of_detected_vehicle.insert(0, right)

@AbdullahKhanML
Copy link

@ChampionTej05 i have changed all the roi position in vehicle_detection_main.py, speed_prediction.py and visualization_utils.py but still cannot counting vehicles, help me please :(
error

Did you find any solution to this, please?

@lizhi501
Copy link

lizhi501 commented Oct 1, 2021

https://drive.google.com/open?id=1i9_K06xxP_k5qBJ7QGFdvKMWwbzcq0J3
This are the listed changes you would need. Roi_detection.py : ROI_POSITION= 1300 (you can set according to your frame size, if your frame size 1920 then set it to 1300) Around 277 line : cv2.line(input_frame, ( ROI_POSITION,0), (ROI_POSITION,height_img), (0, 0, 0xFF), 5) Visualization_utils.py: ROI_POSITION: should be same as above Around 181: Change the IF condition to : left < ROI_POSITION Speed_Prediction.py: Roi_position: Same as above Changes around line 39: many changes hence I am attaching file for it. I have attached other files also. You can check. Vehicle_tensorflow

Hi @ChampionTej05, I was just wondering if you are still able to share the files what what else needs to change around line 39 in the speed_prediction.py file to change the ROI. I'm having trouble getting it to count vehicles when I change the ROI (either on the sample video, or on my own video).

Many thanks for all your help with this, you've done so much for people so far!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests