Even though your project has a problem which is it has nothing to do with Haar classifiers, it turned out to be working efficiently while running real-time detection with passable accuracy. It basically couldn’t detect cars in any given parking lot and all sort of objects that might happen to have a significant amount of edges will automatically change the status of the contours when applying Laplacian edge detection. Anw, Keep up the good work mate!
I am going to publish a further explanation and more tweaked version of this code. Check out ThinYoloV3 and try to make this system Real-time with it's 34-40fps rate with good accuracy.
Thank you so much Ankit. This is a great project. This helps for my university assignment also. I learned lot of things from you. I tried this one.It's working. But my output video is processing very slowly. Why is that? Please tell me a solution. Big help 🙏🙏
Think of the time complexity. Think of how much data is being processed in one iteration and how your processor and GPU is doing it. This would give you the answers.
May I ask why is it that when I run the click_and_crop function, my X and Y coordinates in the Khare_yml_02.yml file are in separate entries instead of being expressed in a list? What am I doing wrong?
There's a script, realtimeboxing, that allows to draw boxes by clicking four points. Those boxes get stored in a yml file and gets used during the runtime
Why you are not making videos for the explanation of other yolo and deep neural network based parking system And according to you which one is the most efficient. Thanks alot in advance 🙏
Hi, please read the medium post of github explanation of the easy street parking. It will clarify most of the things. In case there's something missing or there's a doubt feel free to ask. I'll try and make videos soon.
@@vodcastwithak i know you use jupiter notebook to run the code but can you update your github with the link of medium post. It will be then easy to communicate with you 😊
@@shubammanhas208 medium.com/@ankitkhare/easy-street-parking-with-mask-rcnn-8a1d57a63317 Exact same content is on my github page within easy street parking repository.
How can i use this script for mobile app? Please can you explain step by step?It's a big help.I want to show dynamically slots availability in mobile app's interface.
just trained the model and successfully checked the occupancy of a single parking slot. Can u tell how to implement it in the entire parking area ? Besides, great video man.
The code runs for all the selected spots that you mark using the script. Those spots gets stored in a yaml file and then picked up by classifier one by one .
Yes. HAAR classifier is not as accurate as a Deep Neural Network based classifier. So, I've transfer learned Resnet101 on a set of cars from Stanford car database and then utilized it under this same project. No retraining would be required for any lot if you use that one.
@@vodcastwithak I have successfully run this code with python 3.7 and latest opencv. I am heading onto Transfer Learning. Thank you for your remarkable work and sharing with us. It'll help me with my final project.
Thanks a lot man! This help a lot! May GOD bless you with happiness and knowledge so that you continue to post cool stuff like this!
Even though your project has a problem which is it has nothing to do with Haar classifiers, it turned out to be working efficiently while running real-time detection with passable accuracy. It basically couldn’t detect cars in any given parking lot and all sort of objects that might happen to have a significant amount of edges will automatically change the status of the contours when applying Laplacian edge detection. Anw, Keep up the good work mate!
I am going to publish a further explanation and more tweaked version of this code. Check out ThinYoloV3 and try to make this system Real-time with it's 34-40fps rate with good accuracy.
Hi Khare, I'm very appreciate your work, it's helping me alot. Thanks you for generously explain this project for us. Many thanks again
Thanks
thank you so much bro, i used your algorithm to do my final year defense
Final year defense in what degree?
Please Send working code rushikeshkoli1000@gmail.com
did you publish the research paper on this project
I am also working on same project for my collge hope you will guide me
Thank you so much Ankit. This is a great project. This helps for my university assignment also. I learned lot of things from you. I tried this one.It's working. But my output video is processing very slowly. Why is that? Please tell me a solution. Big help 🙏🙏
Think of the time complexity. Think of how much data is being processed in one iteration and how your processor and GPU is doing it. This would give you the answers.
May I ask why is it that when I run the click_and_crop function, my X and Y coordinates in the Khare_yml_02.yml file are in separate entries instead of being expressed in a list? What am I doing wrong?
Hi Ankit ….I couldn't run the first code show me error ....
pls help me out ...
ValueError Traceback (most recent call last)
in
150 # cv2.imshow('dss',bw)
151 # cv2.imwrite("frame%d.jpg" % co, bw)
--> 152 (_, cnts, _) = cv2.findContours(bw.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
153 # loop over the contours
154 for c in cnts:
ValueError: not enough values to unpack (expected 3, got 2)
How to fix it? :(((
What version of OpenCV do you had? The code doesn´t work in the last version...
Did you predefined the parking spots or is there any code you wrote to draw those polygons at parking spots?
There's a script, realtimeboxing, that allows to draw boxes by clicking four points. Those boxes get stored in a yml file and gets used during the runtime
@@vodcastwithak Can you share the link?
@@sainimohit23 it's all in the git repo.
Do we need to train the classifier to identify cars each time we change the video or use a live feed??
Can we save to text
Like
Parkingslot id parking status
Sir actually I have been working on the same project but there are errors which I am unable to fix so if you could help me it would be a great help
Bro i have worked a bit on your project, can i publish a paper on it in my own words?
Why you are not making videos for the explanation of other yolo and deep neural network based parking system
And according to you which one is the most efficient. Thanks alot in advance 🙏
Hi, please read the medium post of github explanation of the easy street parking. It will clarify most of the things. In case there's something missing or there's a doubt feel free to ask. I'll try and make videos soon.
@@vodcastwithak i know you use jupiter notebook to run the code but can you update your github with the link of medium post. It will be then easy to communicate with you 😊
@@shubammanhas208 medium.com/@ankitkhare/easy-street-parking-with-mask-rcnn-8a1d57a63317
Exact same content is on my github page within easy street parking repository.
Hello, did you manually or automatically create the boxes? How did you detect the parking space covered by a car in the front? Thanks
they are created manually as described in the video.
In this project, it is a script that enables you to make boxes by clicking the four corners
How can i use this script for mobile app? Please can you explain step by step?It's a big help.I want to show dynamically slots availability in mobile app's interface.
Outstanding work. Where can you see the publication for reference?
픽셀변화에 따른 검출이 아니 딥러닝을 사용한 것인가요?
U r god macha👑
It's a typo and he meant, "*good"
Brother can you please help me through this project, there are errors, if you can get me this project working maybe I can pay you
just trained the model and successfully checked the occupancy of a single parking slot. Can u tell how to implement it in the entire parking area ? Besides, great video man.
The code runs for all the selected spots that you mark using the script. Those spots gets stored in a yaml file and then picked up by classifier one by one .
Why are you not putting more videos like this:'(
i am not able to do real time boxing .what to do?
Where is the demo video ??
will it work for if we give another parking lot video file?
yes
Great project.
which course is to be learn to understand this project completely??
Python programming, a little bit of computer vision sprinkled with understanding of what a classifier is.
When we switch to a new parking lot, or we change the camera perspective, I guess we have to relabel and retrain, which is quite cumbersome.
Yes. HAAR classifier is not as accurate as a Deep Neural Network based classifier. So, I've transfer learned Resnet101 on a set of cars from Stanford car database and then utilized it under this same project. No retraining would be required for any lot if you use that one.
Bro how to classify parking spaces in open area
Ankit can you tell me which python and opencv version you have used.
Not sure, should be python 2.7 and opencv 3.2
@@vodcastwithak I have successfully run this code with python 3.7 and latest opencv.
I am heading onto Transfer Learning. Thank you for your remarkable work and sharing with us. It'll help me with my final project.
hello, may i know te exact hardware you chosed for your project?
Inspiron 13 7000 2-in-1 Laptop. You can check its specs.
@@vodcastwithak i mean the camera, you used the laptop camera?
Please share the idea of how to implement it ...please explain asap
that's what the video gives. A high level idea of how to implement it. You even have the code on my github to see exactly what's being done.
Did you improve accuracy in this model?
Yes, as described in the GitHub repo, using deep learning-based classifier improved accuracy.
Can u explain more id assigning part.
Nice video 👍 u deserve more 👌
Can you share me your code for guidance...i really need for my project.
you can check it on my git.
thank you bro
great project!!
ValueError Traceback (most recent call last)
in
150 # cv2.imshow('dss',bw)
151 # cv2.imwrite("frame%d.jpg" % co,bw)
--> 152 (_, cnts, _) = cv2.findContours(bw.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
153 # loop over the contours
154 for c in cnts:
ValueError: not enough values to unpack (expected 3, got 2)
ValueError Traceback (most recent call last)
in
150 # cv2.imshow('dss',bw)
151 # cv2.imwrite("frame%d.jpg" % co, bw)
--> 152 (_, cnts,_) = cv2.findContours(bw.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
153 # loop over the contours
154 for c in cnts:
ValueError: not enough values to unpack (expected 3, got 2)
any possible help