leonlok / multi-camera-live-object-tracking Goto Github PK

I am getting the following error when attempting to run this application.

Any ideas as to what is going on (what I am doing wrong)?

• Serving Flask app "app" (lazy loading)
• Environment: production
  �[31m WARNING: This is a development server. Do not use it in a production deployment.�[0m
  �[2m Use a production WSGI server instead.�[0m
• Debug mode: off
• Running on http://0.0.0.0:5000/ (Press CTRL+C to quit)
  127.0.0.1 - - [11/Mar/2021 16:23:21] "�[37mGET / HTTP/1.1�[0m" 200 -
  Using TensorFlow backend.
  Starting YOLO thread for device 0.Starting YOLO thread for device 1.

Stopping YOLO thread for device 0 due to error.Stopping YOLO thread for device 1 due to error.
[2021-03-11 16:23:23,842] ERROR in app: Exception on /video_feed/yolo/0 [GET]
Traceback (most recent call last):
File "/home/user/.local/lib/python3.8/site-packages/flask/app.py", line 2447, in wsgi_app
response = self.full_dispatch_request()
File "/home/user/.local/lib/python3.8/site-packages/flask/app.py", line 1952, in full_dispatch_request
rv = self.handle_user_exception(e)
File "/home/user/.local/lib/python3.8/site-packages/flask/app.py", line 1821, in handle_user_exception
reraise(exc_type, exc_value, tb)
File "/home/user/.local/lib/python3.8/site-packages/flask/_compat.py", line 39, in reraise
raise value
File "/home/user/.local/lib/python3.8/site-packages/flask/app.py", line 1950, in full_dispatch_request
rv = self.dispatch_request()
File "/home/user/.local/lib/python3.8/site-packages/flask/app.py", line 1936, in dispatch_request
return self.view_functionsrule.endpoint
File "/home/user/Downloads/Multi-Camera-Live-Object-Tracking-master/traffic_counting/app.py", line 60, in video_feed
return Response(gen(camera_stream=camera_stream(feed_type, device, port_list), feed_type=feed_type, device=device),
File "/home/user/Downloads/Multi-Camera-Live-Object-Tracking-master/traffic_counting/camera_yolo.py", line 24, in init
super(Camera, self).init(feed_type, device, port_list)
File "/home/user/Downloads/Multi-Camera-Live-Object-Tracking-master/traffic_counting/base_camera.py", line 78, in init
while self.get_frame(self.unique_name) is None:
File "/home/user/Downloads/Multi-Camera-Live-Object-Tracking-master/traffic_counting/base_camera.py", line 90, in get_frame
return BaseCamera.frame[unique_name]
KeyError: ('yolo', '0')

module 'tensorflow' has no attribute 'Session'127.0.0.1 - - [11/Mar/2021 16:23:23] "�[1m�[35mGET /video_feed/yolo/0 HTTP/1.1�[0m" 500 -
[2021-03-11 16:23:23,843] ERROR in app: Exception on /video_feed/yolo/1 [GET]
Traceback (most recent call last):
File "/home/user/.local/lib/python3.8/site-packages/flask/app.py", line 2447, in wsgi_app
response = self.full_dispatch_request()
File "/home/user/.local/lib/python3.8/site-packages/flask/app.py", line 1952, in full_dispatch_request
rv = self.handle_user_exception(e)
File "/home/user/.local/lib/python3.8/site-packages/flask/app.py", line 1821, in handle_user_exception
reraise(exc_type, exc_value, tb)
File "/home/user/.local/lib/python3.8/site-packages/flask/_compat.py", line 39, in reraise
raise value
File "/home/user/.local/lib/python3.8/site-packages/flask/app.py", line 1950, in full_dispatch_request
rv = self.dispatch_request()
File "/home/user/.local/lib/python3.8/site-packages/flask/app.py", line 1936, in dispatch_request
return self.view_functionsrule.endpoint
File "/home/user/Downloads/Multi-Camera-Live-Object-Tracking-master/traffic_counting/app.py", line 60, in video_feed
return Response(gen(camera_stream=camera_stream(feed_type, device, port_list), feed_type=feed_type, device=device),
File "/home/user/Downloads/Multi-Camera-Live-Object-Tracking-master/traffic_counting/camera_yolo.py", line 24, in init
super(Camera, self).init(feed_type, device, port_list)
File "/home/user/Downloads/Multi-Camera-Live-Object-Tracking-master/traffic_counting/base_camera.py", line 78, in init
while self.get_frame(self.unique_name) is None:
File "/home/user/Downloads/Multi-Camera-Live-Object-Tracking-master/traffic_counting/base_camera.py", line 90, in get_frame
return BaseCamera.frame[unique_name]
KeyError: ('yolo', '1')
module 'tensorflow' has no attribute 'Session'
127.0.0.1 - - [11/Mar/2021 16:23:23] "�[1m�[35mGET /video_feed/yolo/1 HTTP/1.1�[0m" 500 -

Dear LeonLok,

Firstly, I would like to commend you on the exceptional work you have done with the Multi-Camera Live Object Tracking repository. The integration of Deep SORT with YOLO v4 and the comprehensive feature set for traffic counting and object tracking are impressive achievements.

Issue Overview

However, I have observed an area where the application could potentially be enhanced. As noted in your README, the DETRAC dataset primarily contains images of traffic in China, which may limit the model's ability to accurately detect and classify vehicles in different geographical contexts. For instance, the misclassification of hatchbacks as SUVs or the inability to detect taxis with varying colour schemes could be attributed to the lack of diverse training data.

Suggested Enhancement

To address this, I propose the inclusion of additional training datasets that encompass a broader variety of vehicle types and traffic scenarios from multiple countries. This could improve the model's generalisation capabilities and detection accuracy across different regions.

Potential Benefits

• Increased Accuracy: More diverse training data can lead to better model performance when deployed in non-Chinese traffic environments.
• Broader Applicability: Enhancing the model's detection capabilities can make the application more versatile and appealing to a global user base.
• Reduced Misclassification: With a more varied dataset, the likelihood of vehicle misclassification can be significantly reduced.

Conclusion

I believe that this enhancement could further solidify the application's utility and accuracy in various traffic contexts. I would be interested to hear your thoughts on this suggestion and whether it aligns with the future development plans for the project.

Thank you for considering this issue. I look forward to your response and am excited about the potential improvements to this already fantastic tool.

Best regards,
yihong1120

The yolo_frames method in camera_yolo.py takes the frames from the ('camera', '0') stream and then applies YOLO and tracking to each frame. These are then used in the yolo_thread method in base_camera.py.
If camera_yolo.py was successful, then the ('yolo', '0) stream should be working. I think maybe something is wrong within these steps somewhere since your camera stream has no issues?
I hope that makes sense - I think I've never had this issue before. Please let me know how it goes.

Thanks a lot, I will try to figure it out. I 'm feeling thankful when fixing bug and got some help

Hello Duy-Cao-Vo
I have been having this same issue for some time now.
Did you ever find a solution to this issue?
That would be greatly appreciated.

Originally posted by @DataCowboy30 in #8 (comment)

Ok, so how long is it meant to be to run the video streamer and traffic counter? Because, YOLO is still examining the first frame after 10 minutes.


Is there something i need to do to improve the time?
Setup: used yolo4.weights, did convert.py on the provided data and trained model (i.e. used the provided model and only did the conversion on tensorflow 1.14) then opened up a local server (Jupyter). Then ran the streams and opened up the IP address.

I used the detrac tools folder to convert the detrac dataset to market1501 format
Now i want to train it and the code in the cosine metric learning repo expects two folders i.e train and test if i am not wrong.
Can i just split the dataset into 75-25?
and also should i change the size of the images from 100*100 to something else?
Any help will be appreciated.

I have used your repo. While running camera_client_0.py, the following error is encountered.
Traceback (most recent call last): File "C:/Users/dhira/Office Program/YoloV4Multcam/object_counting/camera_client_0.py", line 16, in <module> sender.send_image(cam_id, frame) File "C:\Users\dhira\anaconda3\envs\tf-gpu\lib\site-packages\imagezmq\imagezmq.py", line 106, in send_image_reqrep if image.flags['C_CONTIGUOUS']: AttributeError: 'NoneType' object has no attribute 'flags'

I followed the issue #2 and added the username and password as follows:
path = "rtsp://admin:password@IPAdress:port/
However, the issue is still unsolved.

P.S. I ran the app.py and it is running, but I still get these
.

Can anyone please help me with this?

Hi , I am able to run cam 0 and yolo0 but while running yolov1 (after changing <!img/> to in templates) i am facing this error screenshot - https://ibb.co/fHxyNZf
"Cannot interpret feed_dict as Tensor : Tensor ("placeholder :0", shape=(3,3,3,32) is a element of graph)" . Could you let me know what i should change

Can you update/ send a link to download the weight you trained on Detrac dataset? Thanks a lot

When running python app.py

I got similar error for both flask apps(object counting/traffic counting):

Stopping YOLO thread for device 0 due to error.
name 'decode' is not defined
[2020-07-17 20:00:03,600] ERROR in app: Exception on /video_feed/yolo/0 [GET]
Traceback (most recent call last):
File "/usr/local/var/pyenv/versions/anaconda3-5.3.1/lib/python3.7/site-packages/flask/app.py", line 2447, in wsgi_app
response = self.full_dispatch_request()
File "/usr/local/var/pyenv/versions/anaconda3-5.3.1/lib/python3.7/site-packages/flask/app.py", line 1952, in full_dispatch_request
rv = self.handle_user_exception(e)
File "/usr/local/var/pyenv/versions/anaconda3-5.3.1/lib/python3.7/site-packages/flask/app.py", line 1821, in handle_user_exception
reraise(exc_type, exc_value, tb)
File "/usr/local/var/pyenv/versions/anaconda3-5.3.1/lib/python3.7/site-packages/flask/_compat.py", line 39, in reraise
raise value
File "/usr/local/var/pyenv/versions/anaconda3-5.3.1/lib/python3.7/site-packages/flask/app.py", line 1950, in full_dispatch_request
rv = self.dispatch_request()
File "/usr/local/var/pyenv/versions/anaconda3-5.3.1/lib/python3.7/site-packages/flask/app.py", line 1936, in dispatch_request
return self.view_functionsrule.endpoint
File "app.py", line 58, in video_feed
return Response(gen(camera_stream=camera_stream(feed_type, device, port_list), feed_type=feed_type, device=device),
File "/Users/tm/Documents/TM/neox/Multi-Camera-Live-Object-Tracking/object_counting/camera_yolo.py", line 22, in init
super(Camera, self).init(feed_type, device, port_list)
File "/Users/tm/Documents/TM/neox/Multi-Camera-Live-Object-Tracking/object_counting/base_camera.py", line 78, in init
while self.get_frame(self.unique_name) is None:
File "/Users/tm/Documents/TM/neox/Multi-Camera-Live-Object-Tracking/object_counting/base_camera.py", line 90, in get_frame
return BaseCamera.frame[unique_name]
KeyError: ('yolo', '0')

Searched other places, no good, seems not general issue.

I'm using:
Tensorflow 1.14
Keras 2.3.1
opencv-python 4.2.0
ImageZMQ
numpy 1.18.2
Flask 1.1.1
pillow

on a MacBook(10.15.5), also tried RasbianOS, both got above error.

Python version I'm using is 3.7.2, the recommended is 3.6, not sure if this is the key.

Thanks.

Hi there,
I love your project, thanks for sharing! Is there a possibility to share your Yolo weights trained on DETRAC? It is missing the model_data/yolov4.h5 now to run it completely

[2021-10-18 16:14:38,775] ERROR in app: Exception on /video_feed/yolo/0 [GET]
Traceback (most recent call last):
File "C:\Users\admin\Anaconda3\envs\new\lib\site-packages\flask\app.py", line 2073, in wsgi_app
response = self.full_dispatch_request()
File "C:\Users\admin\Anaconda3\envs\new\lib\site-packages\flask\app.py", line 1518, in full_dispatch_request
rv = self.handle_user_exception(e)
File "C:\Users\admin\Anaconda3\envs\new\lib\site-packages\flask\app.py", line 1516, in full_dispatch_request
rv = self.dispatch_request()
File "C:\Users\admin\Anaconda3\envs\new\lib\site-packages\flask\app.py", line 1502, in dispatch_request
return self.ensure_sync(self.view_functions[rule.endpoint])(**req.view_args)
File "app.py", line 57, in video_feed
camera_stream = import_module('camera_yolo').Camera
File "C:\Users\admin\Anaconda3\envs\new\lib\importlib_init_.py", line 126, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
File "", line 994, in _gcd_import
File "", line 971, in _find_and_load
File "", line 955, in _find_and_load_unlocked
File "", line 665, in _load_unlocked
File "", line 678, in exec_module
File "", line 219, in _call_with_frames_removed
File "C:\Users\admin\Desktop\boats\new\Multi-Camera-Live-Object-Tracking\object_counting\camera_yolo.py", line 9, in
from yolo import YOLO
File "C:\Users\admin\Desktop\boats\new\Multi-Camera-Live-Object-Tracking\object_counting\yolo.py", line 12, in
from yolo4.model import yolo_eval, Mish
File "C:\Users\admin\Desktop\boats\new\Multi-Camera-Live-Object-Tracking\object_counting\yolo4\model.py", line 11, in
from keras.layers.normalization import BatchNormalization
ImportError: cannot import name 'BatchNormalization'

what is the prob, please?

File "camera_client.py", line 16, in
sender.send_image(cam_id, frame)
File "/home/dell/Documents/Prasad_AI/Project3/Flask-Multi-Camera-Streaming-and-Live-Customer-Counting-with-YOLOv3-and-Deep-SORT/myenv/lib/python3.6/site-packages/imagezmq/imagezmq.py", line 106, in send_image_reqrep
if image.flags['C_CONTIGUOUS']:
AttributeError: 'NoneType' object has no attribute 'flags'

when I am running Camera_client.py , I am facing that error , pease help me

Hello!
First, thank you for this impressive work!
However, I read the code, but I do not quite understand when doing MTMC tracking how it makes sure that the same person always gets the same ID across cameras, which is quite important in multi-camera tracking. For instance, if the target numbered 0 appears under Camera 1 and Camera 2 at the same time, how to make sure this constraint is not violated?

Hope you could explain this.

Hi @LeonLok ,
Which IDE are you using?
Best regards,
PeterPham

Hi,

I'm posting this issue because I observed that I get worse scores with a DeepSORT tracker compared to what I get with a simple SORT tracker. I used your DETRAC tool to create the training images for the cosine metric learning. I also tried to re-train my DeepSORT model but I got equivalent results.

I compared the HOTA scores obtained on several cameras of AICity Challenge (Track 3) using the same detector (YOLO v4). I tried to tweak the max_consine_distance parameter of nn_matching and it seems that decreasing the distance increases the score. I obtained the best result when I set it to 0, I guess it means that I'm much closer to SORT in this case. For instance, on one camera, I obtained those scores:

There's not much things to tweak in the training parameters of the cosine metric. However, I was forced to use the triplet loss though, like this issue reports , did you have to use the triplet loss as well? At first, I guessed it was related to the dataset, but now I wonder if this may be the reason why my model does not work very well.

Thank you in advance!

Hello, I do n’t have a webcam IP address, can I load offline videos? How to set?

In the camera_client_0.py,path = "rtsp://192.168.1.70:8080//h264_ulaw.sdp".
How should I replace the entire path to my own camera.

I am not getting YOLO stream.
I got the camera stream but I cannot get YOLO stream. The image can be seen here

The error is as in the following image

Kindly help me, please.

I am able to run everything successfully(till yolo4.h5 is loaded) but nothing appears in browser . You can see the screenshot here https://ibb.co/wc0dfZ2

In readme,you have mentioned that each object is being counted once,how do you ensure that the object is counted only once,what if that same object was viewed from a different camera,wouldn't that result in being counted twice.

As the title，how to set the path such as "rtsp://192.168.179.33:8588//h264_ulaw.sdp"

Hi Sir,
Thanks for this amazing work. I am using your Yolov4+DeepSort project but the speed is very slow. Is it possible to combine Tiny_Yolov4+DeepSort? Where we should make changes for this work in the Yolo and tracker? Do you have any plans to build this project?

Excellent project. In your project do you add re-identification task ?

Have you published a paper that will be used as a reference paper for your paper?
If not, I just want to refer to this github page.
Thanks for the inspiration.

I'm trying to install the environment file but I think there's some compatibility issues due to the latest Anaconda distribution. Could you tell me please which version are you using???

Hi,

First of all, cogratulations on your project. I'm trying to run it streaming my android camera. However, after running both files and opening localhost:5000, this error pops up in the app.py console. I've already modified it to use just one camera. GPU is enabled and already increased the time in base_camera.py to prevent YOLO threat from shutting down. Did you face something similar? I would appretiate your advice.

Stopping YOLO thread for device 0 due to error.
Unable to open file (unable to open file: name = 'model_data/yolo4.h5', errno = 2, error message = 'No such file or directory', flags = 0, o_flags = 0)

Thank you in advance,
-Alex

hello, is there a way to use multiple usb cameras instead of streaming on ip cameras.
thank you.

the ua-detrac dataset tool is amazing, but I got a problem while training, we know that it is max_label = 1501 for market1501 dataset, but what is the max_label for ua-detrac?

File "/home/dell/Documents/Prasad_AI/Project3/Flask-Multi-Camera-Streaming-and-Live-Customer-Counting-with-YOLOv3-and-Deep-SORT/myenv/lib/python3.6/site-packages/flask/app.py", line 2447, in wsgi_app
response = self.full_dispatch_request()
File "/home/dell/Documents/Prasad_AI/Project3/Flask-Multi-Camera-Streaming-and-Live-Customer-Counting-with-YOLOv3-and-Deep-SORT/myenv/lib/python3.6/site-packages/flask/app.py", line 1952, in full_dispatch_request
rv = self.handle_user_exception(e)
File "/home/dell/Documents/Prasad_AI/Project3/Flask-Multi-Camera-Streaming-and-Live-Customer-Counting-with-YOLOv3-and-Deep-SORT/myenv/lib/python3.6/site-packages/flask/app.py", line 1821, in handle_user_exception
reraise(exc_type, exc_value, tb)
File "/home/dell/Documents/Prasad_AI/Project3/Flask-Multi-Camera-Streaming-and-Live-Customer-Counting-with-YOLOv3-and-Deep-SORT/myenv/lib/python3.6/site-packages/flask/_compat.py", line 39, in reraise
raise value
File "/home/dell/Documents/Prasad_AI/Project3/Flask-Multi-Camera-Streaming-and-Live-Customer-Counting-with-YOLOv3-and-Deep-SORT/myenv/lib/python3.6/site-packages/flask/app.py", line 1950, in full_dispatch_request
rv = self.dispatch_request()
File "/home/dell/Documents/Prasad_AI/Project3/Flask-Multi-Camera-Streaming-and-Live-Customer-Counting-with-YOLOv3-and-Deep-SORT/myenv/lib/python3.6/site-packages/flask/app.py", line 1936, in dispatch_request
return self.view_functionsrule.endpoint
File "app.py", line 41, in video_feed
camera_stream = import_module('camera_yolo').Camera
File "/home/dell/Documents/Prasad_AI/Project3/Flask-Multi-Camera-Streaming-and-Live-Customer-Counting-with-YOLOv3-and-Deep-SORT/camera_yolo.py", line 19, in init
super(Camera, self).init(feed_type, device, port_list)
File "/home/dell/Documents/Prasad_AI/Project3/Flask-Multi-Camera-Streaming-and-Live-Customer-Counting-with-YOLOv3-and-Deep-SORT/base_camera.py", line 80, in init
# wait until frames are available
File "/home/dell/Documents/Prasad_AI/Project3/Flask-Multi-Camera-Streaming-and-Live-Customer-Counting-with-YOLOv3-and-Deep-SORT/base_camera.py", line 92, in get_frame
BaseCamera.event[unique_name].clear()
KeyError: ('yolo', '1')
Help me.

[ WARN:0] global /io/opencv/modules/videoio/src/cap_v4l.cpp (887) open VIDEOIO(V4L2:/dev/video0): can't open camera by index
Traceback (most recent call last):
File "camera_client.py", line 11, in
sender = imagezmq.ImageSender(connect_to='http://localhost:5000/')
File "/home/dell/Documents/Prasad_AI/Project3/Flask-Multi-Camera-Streaming-and-Live-Customer-Counting-with-YOLOv3-and-Deep-SORT/myenv/lib/python3.6/site-packages/imagezmq/imagezmq.py", line 50, in init
self.init_reqrep(connect_to)
File "/home/dell/Documents/Prasad_AI/Project3/Flask-Multi-Camera-Streaming-and-Live-Customer-Counting-with-YOLOv3-and-Deep-SORT/myenv/lib/python3.6/site-packages/imagezmq/imagezmq.py", line 62, in init_reqrep
self.zmq_socket.connect(address)
File "zmq/backend/cython/socket.pyx", line 583, in zmq.backend.cython.socket.Socket.connect
File "zmq/backend/cython/checkrc.pxd", line 26, in zmq.backend.cython.checkrc._check_rc
zmq.error.ZMQError: Protocol not supported

I believe is only for OsX users?? Is there a way to run this projects on ubuntu 18.04?

Hi LeonLok your work is really inspiring, I'm somewhat of a beginner and i'd like to know how or where you are able to determine the FPS on yolov3 and yolov4 when you run the model on videos.
Thank you

it would be nice if u can modify the code for offline video version

Hello,

Thanks for the sharing this wonderful project. I have a quick question. Is this running seperate trackers for each stream? What I mean is, let's say an object is given an id of 10 in camera1, then the object leaves camera1 and enters camera2, does it get the same id of 10? If not, is it possible to achieve something like this?

Kind regards,
Burak

hi ...
which IDE are you using?
Best regards,
PeterPham

To test out the app i found some open source stream rtsp://wowzaec2demo.streamlock.net/vod/mp4:BigBuckBunny_115k.mov
The i run app.py and than camera_client_0.py the flask app shows in logs
Cannot feed value of shape (1, 160, 224, 3) for Tensor 'input_1:0', which has shape '(?, 608, 608, 3)'

Does anybody faced this issue?

can you please explain what this variable (max_track_id) does? By the time crop_dataset creates images, this variable has a value about 6000 and the cosine metric learning in nwojke's repo throws me an error saying the id is greater than 1502