Counting Objects Using YOLOv4 Object Detection | Custom YOLOv4 Functions with TensorFlow
ฝัง
- เผยแพร่เมื่อ 5 ต.ค. 2024
- Learn how to code your very own Custom Functions to work with YOLOv4 Object Detections! In this video I will walk-through how to run an object counting app using YOLOv4, YOLOv4-tiny, YOLOv3 or YOLOv3-tiny. The commands can be run using TensorFlow 2.0, TensorFlow Lite or TensorRT models on images, video and webcam!
#yolov4 #objectdetection #tensorflow
This video will cover two custom functions that have been created with YOLOv4. The first being for counting objects within images and video, and the second being a custom flag to show detailed information on YOLOv4 detections (class, confidence score, bounding box coordinates).
GET THE CODE HERE: github.com/the...
Follow the tutorial to be able to convert YOLO style Darknet weights into saved TensorFlow models, and run these models. Using easy command line flags to enable custom functions like object counting. Take advantage of YOLOv4 as a TensorFlow Lite model, it's small lightweight size makes it perfect for mobile and edge devices such as a raspberry pi. Looking to harness the full powers of a GPU? Then run YOLOv4 with TensorFlow TensorRT to increase performance by up to 8x times.
In this video I cover:
1. Show Custom Functions Repository and Explain Object Counting
2. Cloning or Downloading the Code
3. Downloading and Converting YOLOv4 Weights into a saved TensorFlow Model
4. Enabling Object Counting while performing YOLOv4 Object Detections with TensorFlow on images, video and webcam
8. Printing Detailed Information About Object Detections to command prompt
9. Stepping through the code to create custom functions
-------------------------Resources--------------------------
Train Your Own YOLOv4 Custom Object Detector in the Cloud: • YOLOv4 in the CLOUD: B...
Running Pre-trained YOLOv4 model with TensorFlow, TFLite, TensorRT: • YOLOv4 Object Detectio...
running Custom YOLOv4 License Plate Detector Model with TensorFlow, TFLite, TensorRT: • How to Build a Custom ...
The Official YOLOv4 paper: arxiv.org/abs/...
If you enjoyed the video, toss it a like! 👍
To Subscribe: / @theaiguy
Thanks so much for watching!
The AI Guy
![ลองสั่งของจากแอพ temu ถูกจริงหรอ?? ( part : 1/2 ) [โกงไหมครับ ep.89 ] | DOM](http://i.ytimg.com/vi/RTcZSQTiDZw/mqdefault.jpg)
Let me know which Custom Functions you want me to build next here in the comments! Cheers :)
i want you to find the distance of detected object from camera. mean distance between camera and detected object,
Haroon Asad thanks for the suggestion! Will look into this!
Hi, thanks for this wonderful video guide as I requested earlier. If possible please make object distance,object speed measurement, video analysis etc with YOLO and OpenCV.
@@arindammondal9364 Will definitely be looking into this for future video! Thanks for the suggestion. All the best.
Hi! How do I tweak the code to detect object only on specific areas (2 ou 3) ? Thanks!
Great add on would be, draw line and count number of people crossing the line.
Himesh Panchal that does sound like a great add on. Thanks for the suggestion! Cheers
@@TheAIGuy By any chance, have you worked upon this one?
Thanks for the informative video.....Happy learning & Coding !!
akshay jaryal thanks for the kind words! Cheers
Amazing video, code is clear and easy to understand. Idea, maybe make a passing lines and count objects that passes that line from same direction? i.e. in and out
Dude you are amazing, and i am pissed cause another channel took your stuff and used it as the easiest way to install yoloV4 and they have more subscribers but u got the videos and the repo and everything, you are amazing keep it up and thank you so much for the help here
amazing as always!!! would you plz upload a Colab tutorial of this? and use pre-trained weights to detect one class? benefit a lot from your channel !!
Thanks for the kind words and suggestions! Will definitely look at making a Colab Notebook to run this repository! Also will be showing how to run on only one or two of the classes within the pre-trained model. Thanks again and stay tuned!
@@TheAIGuy hello! I want to use the pretrained weights but only to detect cars. I modified coco.names and obj.data, but it's not working. Is there a way to do this?
Sir, you are great...Really helpful to learn yolov4
Please make a speed detection of objects(for cars trucks bikes etc) and line crossing counter on road that could be helpful to do tracking on vehicles which breaks traffic rules
Great video. Has anyone else having problems where the detection's only work for the first frame?
Amazing Tutorial🤩. Looking for more amazing videos!
Thank you very much for the very good tutorial. May you prepare a similar kind of tutorial for Faster R-CNN?
Great video once again and very informative ! The code walkthrough was very nice.
About the video detection console output, would it be easy to add when an object was found in the video? For example the output would be : Object found at 01:23:433 : dog ...
Thanks a lot :)
Hi, I want to count th enumber of objects in a continuous video feed of camera and store it in real time. How do i do that? for example, if theres a rowof straawberries, as i move the camera, i want to count the number of strawberries it has detected without counting the same strawberry twice. Thanks in advance.
Hey akshay, i want same conditions , are you done that??
@@karthickraja1026 i also need to count cars passed.. using a ROI line.. can someone help
not too bad videos. sometimes they even useful. but bell sounds are too loud(about subscription part). just make them a little bit quiet.
Thank you for the tutorials. Can you apply this process to the tensorflow colab as well?
great tutorial very intuitive i am waiting desperatly for the DeepSort tracking with yolov4.
You are still my go to for learning AI. AWESOME!!
YOLO V4 with deepSORT like your tutorial in YOLO V3? plus output csv file of number of object detected, class name, and prediction score please.
Thanks for the kind words and suggestions! DeepSORT with YOLOv4 is my next video and will be out next week! So stay tuned.
Thanks for the video
I have a task which is pedestrian detection and congestion measurement
I need to build a model that can detect pedestrian and also be able to say if the frame is congested
For example let's say in a frame greater than 20 people means congested
Thanks for the video. It was really useful. Can you please make a tutorial to count no of chicks in a video by using yolo
great work as always😍🤩.can i ask if is it possible to use the pre-trained weights to detect 1 class only?cuz you not intrested in the other 79
Yes this is an easy change, so I will for sure add this to the code and show it in a future video! Thanks for the suggestion!
@@TheAIGuy thank you so mcuh
Great stuff here! Highly recommended. Thank you
Nice!
Love it!! and I love the issues system!!
Hi AI GUY great videos , can you do a video on object tracking using multiple camera at a time , i tried to search about it but failed, if possible please make a tutorial on this or give any good source plzz
- A subscribed person
Can we detect multiple images in 1 folder, not doing it with one by one images?
How to crop amd store image of the detected object?
Love your videos btw!
I’m getting an error and shows
ValueError: cannot reshape array of size 2126904 into shape (1024,512,3,3)
I’m using a custom yolov4 trained weight with obj.names containing only a single class
Correction made in config.py
Still showing this error, tried with a yolov3 custom trained weight and also same situation
Can we do this using TensorFlow Lite?
great video, have you done something similar in Yolov5
I'm trying to build object detection on solar system's planet but I don't know where to get dataset for that (planet detection)
just hard code it! XD
this is a great video it's really helpful, thanks for u'r hard work .So I just want to ask how I can get the number of people class not on the screen but as integer after implementation of this algorithm on image not on video
Nice video
Please can this counting system be applied to YOLO v5??
great tutorial!, thanks a lot for sharing !
You are really great, I learned yolo from your channel... Do you have any plan for faster RCNN and Mask RCNN
Hi there,
First of all, thx for your work. It is a nice program.
I have a question about the counter in the project.
If I want to display the total number of detections and the number of classification detections, instead of showing the classification number and the total number of detections in the current frame. How do I change the function.py?
The first time, I dele the 'else' in the count. But only the items detected in the current frame can be displayed, and the sum function cannot be realized.
Can u give me any advice?
Thanks again for your help! Is there any function that allows us to get the label of the objects detected as an output, in order to use them in an app?
Thanks and great content
Great video. I really like the ease of installation in your tutorials with the conda environments. Do you know how it is possible to do convert the tflite models for yolov4 tiny and yolov4 to run them on a coral edge tpu? Also it would be cool to have a function to reconstruct an image with bounding boxes by combining the bounding boxes coordinates you print with the --info tag and the original image. What do you mean by the way with bounding box coordinates relative to the image. Does xmin = 130 and xmax = 140 mean, the bounding box goes from pixel number 130 to pixel number 140 in the original image or how can we exactly interpret that values?
hi Bro, thanks for the awesome tutorial, i need your little help in putting ROI lines also to count the total cars passed that line.
please responsed.
Thanks
Hello nice tutorial video. I want to ask can the counter accumulate the number of object counted?
Cool
can you make a video on solar panel detection and count ?
hello and I smash subscribe button.
super olo you the best! :)
Any ideas about why I got the warning during darknet to TF conversion when I use Tensorflow GPU 2.6.0? The warning is like this: WARNING:TensorFlow:Compiled the loaded model, but the compiled metrics have yet to be built. model.compile_metrics will be empty until you train or evaluate the model. After that when I try to do inference multiple images, the bbox result only appears only on the first image, and not the rest of them. Same with the video, only inference the first frame. Any help would be appreciated 🙏.
GREAT VIDEO!!!
Can we make the detect.py faster?? This detect.py seems to be slower than the your previous video
I would like to see object tracker using YOLOv4, unless its the same as YOLOv3 and I am dumb
Arhanel I will be doing an object tracker with Yolov4 shortly in a future video. It is not quite the same as yolov3!
@@TheAIGuy Oh, good to know. I was trying to create it myself for a project I am working on currently (internships and stuff like that) but I am new in object detection. There are so many things to consider, sometimes its hard. I managed to do some cool stuff using your YOLOv3 tracker tutorial but most of the time it misses objects and/or loses tracks + its slow, but this doesn't matter. Was thinking about training my own model, because I want only to track cars, truck and peoples but after 3000 iterations it did nothing, so I guess I need more research. Thank you for your tutorials, great job.
Nice tutorial, i have a question, why do you don´t used darknet whit yolo?. Regards!
Seba Segovia I use darknet to train models. But when it comes to creating application with Yolov4 i find it a lot more flexible and easy to customize the code when running it with tensorflow.
hola seba, estas en argetina?
thanks for this video amazing , my manager ordering me make counter our divaces more the 400 divaces like printer - computer - mouse - table - chare - laptop etc ) can i make it use this artcale then save the resulit to database or excelsheet
hey, how to change the number of classes to 1 i.e just to display the number of persons?
Hello
Thank you for the video very useful ! I was wondering whether it was possible to count multiple images at once and then save the outputs as pictures and text file.
Bro do you success to run it? Cause I need it too
Hey bro nice video , I just wanted to ask if we can create a function to count detected objects only within a detection boundary , if so how?
Great job AI. please make a video on deep SORT tracker with yolov4. Can I use custom v4 weights file accordingly to your deepsort yolov3 tracking tutorial? I mean change custom v4 into tf files and run tracker
Hello sir can I use the function counting object directly to the Yolo v5 Instance segmentation? thankyou subscribing
Hi @The AI Guy! This is a great video and really helpful!
When I run this for a video, it's only getting the counts for the first frame only. For other frames not detecting anything. Number of total objects getting zeros after the first frame :(
Can anyone help me with this?
Any help would be much appreciated.
Hi, Really interesting and helpful video. Can you please guide me: how to get real-time graphical variation continuously as the X-axis (average people Number) and Y-axis (Average height) of each frame running as video processing. I want the animated real-time plot as a video running continuously. Thank you and Cheers..
Can we use it in google colab? "--dont_show" is not working. Its shows cannot connect to X server
MEET MOTA will look into this and i will test it more!
To bypass "cannot connect to server X" error, comment CV functions that want to open a new window:
in detect_video.py, comment
"#cv2.namedWindow()
#cv2.destroyAllWindows()
@@tinyml_blog6012 I have tried this. However, it works for the first frame only. It no longer detects afterwards
Have you fixed the above error ?
#cv2.namedWindow()
#cv2.destroyAllWindows()
I commented out these 2 line of codes. This worked for me.
How can we use this in google colab?
How do I get a hold of you ? Would like you to help me on a project.
Thanks
Can you tell what needs to done, if we want to display every object count sequentially inside its corresponding bbox.
can anyone tell me how i can take all this and make a standalone program, like customize windows and stuff for like end product
Hi @ The AI GUY getting an error
Traceback (most recent call last):
File "save_model.py", line 4, in
from core.yolov4 import YOLO, decode, filter_boxes
ModuleNotFoundError: No module named 'core.yolov4'
managed to find the solution after a while. For some reason python does not read folders, so u need to copy the files in the core file and edit all scripts removing the core on the core.file. It works fine for other pc, tested it on a gpu without error. Maybe because I was using a cpu machine. Hope it helps.
Súper !
How Can we use two different models in one session, for example, one model to identify kitchen utensils and different models to identify the knife . How can we merge those 2 models in one? can you please help.
Thanks In advance
If I want to run it in the colab
Is it also working with Yolov5? I need to count 2 classes of objects an upload the value in a database.
I highly recommend not using Yolov5 as it has been debunked as fake. The authors of it made small changes to yolov3 and then called it yolov5. Yolov4 is way better than yolov5 so i recommend yolov4.0
As I am working on hand tracking. and i want to plot the centroid of both hands. How i can do this using yolo
I'm wondering if we can apply this uing an SSD Mobilenet Inception v2 Model.
It detects fine in realtime but reverses left and right in the output video. Please help!
So how much do we lose in accuracy when we use the tiny weights ?
Is there a way to capture the image after the detection occur? Thanks in advance.
Is there any option to get coordinates of the object detected in image. Have used yolov4 but unable to get coordinates
I need help in my final year project they give me a vehicle detection project for student and staff vehicles only how to do?
are there any documentation to implementing this on a matlab program?
awesome,,,,,,,,,
How to count the total objects in the video and keep accumulating .i really appreciate.
This can be implemented by having a global counter. Currently, it is coded to reset with each new frame. By initializing the counter outside of the while loop you should be able to get this to work. Hope this helps!
@@TheAIGuy Thanks so much!! I will try it later. I really enjoy your videos. Thanks again!!
@@TheAIGuy Won't this add the count on each frame to the previous frame's count? so if there are 25 frames of one person walking down the street, it will return 25 persons as count? by simply initializing the counter outside the while loop, how would YOLO know whether the person is the same person as the previous frame?
How could you count all the objects throughout a video, without double counting?
. I am building a video monitoring application locally. Is it possible to write one code that will work for android, ios and desktop?
can you post atutorial how you write this code from scratch ?
I got this error
F .\tensorflow/core/kernels/random_op_gpu.h:232] Non-OK-status: GpuLaunchKernel(FillPhiloxRandomKernelLaunch, num_blocks, block_size, 0, d.stream(), gen, data, size, dist) status: Internal: no kernel image is available for execution on the device
Fatal Python error: Aborted
I have the same error, have you found any solution?
@@cristiandavidparramunoz6101 use TF 2.2 GPU version
Helo bro, I need ur help,we are working on a project in which we need a vehicle and pedestrian detector boths so would you help me?
how to train on custom dataset on this tensorflow version of yolov4
Can you please tell how to count just persons?
can we do this using yolo(v3 or v4) on jetson nano?
is it possible to run this on raspberry?
Hi, Can you help me create a function that will input multiple images at once and give out the total number of the class detected in the output. I'm running my yolov4 in darknet so can you help me run it in darknet framework. Please it's for a project and urgent.
Thank you!
I've got this error:
no kernel image is available for execution on the device Fatal Python error: Aborted
Can this work with Yolov5?
That code can use in google colab?
i have a question,
cv2.error: OpenCV(4.6.0) :-1: error: (-5:Bad argument) in function 'rectangle'
can anyone help me with this?
can you build this project in android ? please make this type of project in android. i am waiting for that very eagerly
zing! zing! amazing! tutorial
I am facing trouble in converting .weights to tflite version in Gitbash.
It is notifying me with "Permission denied"
I have used this command:
python save_model.py --weights ./data/yolov4.weights --output ./checkpoints/yolov4-416 --input_size 416 --model yolov4 --framework tflite
Please help me!!
Real time object counting how is the possible
hello, i dont know which file i've to mod or how to use your "Count Objects per Class" in Google Colab. ill appreciate any advice. thanks
did you got it how to count labels of particular class ?
vehicle speed estimation
how to detect count of particular object in video using yolov4 reply me for this
i have problem when i run
# Convert darknet weights to tensorflow
## yolov4
python save_model.py --weights ./data/yolov4.weights --output ./checkpoints/yolov4-416 --input_size 416 --model yolov4 . It shows
File "C:\Users\User\anaconda3\envs\yolov4-gpu\lib\site-packages\typing_extensions.py", line 143, in _collect_type_vars
not isinstance(t, _UnpackAlias)
NameError: name '_UnpackAlias' is not defined. How to fix it?
Do you know how can I implement it on android studio? Or if it's possible to implement there?