Count people with a high accuracy | CCTV camera, Opencv and Deep Learning
HTML-код
- Опубликовано: 4 фев 2025
- 👉 AI Vision Courses + Community → www.skool.com/...
Blog: pysource.com/2...
We will see how to identify and count people from CCTV cameras with the highest accuracy possible. I will explain the main problems you might find in making this project and what the best strategy is.
➤ Courses:
Full Computer Vision course: pysource.com/o...
Training Mask R-CNN PRO (Notebook + Mini-Course): pysource.com/m...
➤ Follow me on:
LinkedIn: / pysource
➤ For business inquiries:
pysource.com/c...
#CCTV #opencv #deeplearning
![[Webinar] How to Build a Modern Agentic System](/img/1.gif)
probably the best chanel for who wants to learn more about computer vision.
Very nice video, sir! Your explanation and passion for the topic are truly inspiring.
I'll definitely be learning from all your free courses. Ty sir! 🙏🙏
Thank you so much~ I appreciate your free courses, and it helped a lot...
Can you share the algorithm and how you deploy the system. That would be helpful for me since I`m currently working out on this kind of system for my Research with Raspberry Pi. Thanks!
if you have any source ,can u share with me
Amazing video, and super clear! Thank you!
Would it be possible to provide the source code? would be very useful
View is awesome.
Brother can i use this camera for barber shop for counting persons
Hi Sergio, great video. my understanding from the blog, you have used YOLO for training. any specific version
Respected Sergio,
Can you make a video how to train the CNN with your own dataset?
I have done already but in Matlab and not in Python.
Thanks.
Regards
Hi, nice video. Do you know how multi-camera object tracking works? Different viewing angles, but labeling the same persons with one label? I am trying to accomplish this but have no idea how to start.
you should stich them with panoramic view and make it look like one single frame. But I don't recommend doing that anyway, it won't be reliable.
@@pysource-com I have already tried that. Stitching doesnt work because the environment is too similar. What i am trying at the moment is, to find duplicate persons in 2 overlapping images (cameras only translated in a line). I think about trying IoU and comparing the Bounding-Boxes of detected images (feature vector of classification) or histograms (not very reliable i think). What do you think about this?
@@lukasb7013 Hi, have you tried person ReID? im currently trying to use it to give the same id to a person that could eventually come back to the frame but in theory should work with multiple frames too
@@Frlinkin1 I am currently trying it. Found out about it some weeks ago. Do you have any suggestions what to use? Currently trying to train a model on the Market-1501 Dataset. But i cant test it on my own data. I just need to get the extracted features for the comparison
@@lukasb7013 @Fabiola Ruiz Have you people sorted out multi-camera? Any success with ReID?
what if a headless person gets into the frame???
btw very useful tutorial, thank you for sharing.
People: n / Disguised People: 1
Hello. Can you record a video about how to install OpenCV with CUDA support? My CPU is running your yoloV4 model at 10 seconds per frame since my GPU is not being used.
Hello Sergio, This is a very useful video.
How do you think about FairMOT compare with deep sort, Thank you!
does it work on night time
How do I get the Realtime video to work on it ?
Great video thank you for sharing
Hi, why you use windows not Linux?
Can u provide the video used in the program?
hi how to do railway obstacle detection with opencv and pyhton
Could you please share github link for human count project (code, dataset, others)
congratulations teacher, could you share the code?
hye that can count people using yolo?
Can u provide the video cctv
Adding yolo to arduino projects next please
Is there a github link for the code in this vid ?
Can we modify the code like if person left the roi, it should be minus from the total count?
Yes, it's possible to do that as well. It's possible understand the direction for example and add + or - to the count if person is going to opposite direction
sorry how can i use google open images to get giraffe images???
Hi, how can count boxes going thru a conveyor belt? And how can I get in touch with you for a company project that I need? Ty
The concept is similar to this video, but it's simpler then counting people:
- boxes detection + tracking
for commercial requests you cant contact me here pysource.com/contact/
Hi sir, where can I contact you?
when I try to rum my code I get this output:
ImportError: cannot import name 'ObjectDetection' from 'object_detection'
Can someone help me?
HI, the object Detection library is a custom implementation only available on the course Object Detection (with Opencv and Deep Learning) pysource.com/object-detection-opencv-deep-learning-video-course/
no source code is provided with this video, but it's only a demonstration.
This need a written PDF!.
Hi Sergio,How is The RTSP Live stream In CV?
I am working on basic project like that , The battery vehicles Tracking In that top of the vehicles they past QR code and The arrow mark image I could Find the Vehicle no using Qr code and Find the Direction That the arrow mark Is show lift means it will going out and Right means it will Coming IN in that project My RTSP link in not run 24/7, but It will run 20-30 minutes how could RTSP link and How to Find the arrow mark direction In Video could you give in any video , Please
Do you work with footfall counting machine??
can you give me the key word to install deep_sort and object_detection in pip module
Deep sort can't be installed with pip. It's an open source library that you can find here on github github.com/nwojke/deep_sort and to be used needs to be integrated with Object Detection.
You did not provide the source code, here, why?
Source codes for more advanced projects about object detection and Tracking are only on my videocourses on pysource.com
On RUclips I release simple codes for educational purpose only, while commercial projects and a more structured learning is provided on the courses.
Superb.....i need more after this counting, you used one line right to separate ROI .....i am building gender classification model in particular ROI i need to count particular person with gender...say for example if he is there in 5frames we have to compare these five frames to finalize his gender if in four frames he May identified as male and in one frame he May identified as female then after 5th frame we have to State that he is male.....so how to do it help me with it
did you find a solution ?
whats the matter of posting video like this with out source code??????????
sir can you provide code..?
I'm sorry that's not possible. I do provide here on youtube and on my blog only basic codes for educational purpose.
More advanced codes are only provided inside the courses I offer, or as a part of commercial projects for companies
Can you provide the code for this sir?
I only share on youtube simple codes for educational purposes but more advanced codes for detecting and tracking are only inside the Object Detection with Opencv and Deep Learning course that you can find on pysource.com
wait, doesnt that gonna make your model overfitting or something?? i just wondering btw hehehehee
Plz reply........