YOLO v7 Object Detection Models: FPS & Object Detection Comparison (All 7 Models)

YOLO v7 object detection model has 7 variants. This video showcases FPS and objects detection accuracy of all the models. There is a tradeoff between speed and accuracy, and this comparison is helpful in deciding the sweet spot. Which model variation works best for your case? Let me know in the comments down below.
---------------------------------------------
Want to discuss more?

#TheCodingBug
---------------------------------------------
► My Tutorials:
○ YOLO v7 Object Detection Complete Tutorial:
---------------------------------------------
---------------------------------------------
DISCLAIMER: Links included in this description might be affiliate links. If you purchase a product or service with the links that I provide, I may receive a small commission. There is no additional charge to you!

[felipewitezak]

God job! Make a compare with others yolos

[furkantoptan]

On which device did you get these fps values?

[andrespereira]

Hi Mr TheCodingBug. I like a lot your channels for all the well explained content that you were creating all this time. Now, back to this video... what is the hardware specs that you used to test the 7 models?

[ZionZhou-ky]

Hello! I'm really curious about the resolution information for these videos. Thank you for your help!

[kin_]

cool video, though i am not sure which is better here?

[MEATHEADBooYA]

Hello again, I have a random question. So I am trying to train on Yolov4 following all your tutorials and its working as intended! However I just cant train with a width and height of 1024.

My settings are

Batch = 64
Sub = 64
Width = 1024
Height = 1024

Running on a 3060 12G. 16G of Ram and 16 core AMD 3650x

I am told I need to have 1024 width and height as that is what the pre-trained / Images are on.

What I get is

1472 x 1472
Create 6 permanent CPU-threads
Error: cuDNN isn't found FWD algo for convolution

I have looked everywhere but I don't see anyone training at 1024 even though that is what my images are on.

What I also notice is that my GPU Memory goes to 11.4G - 11.6G of the 12G and then the Error appears.

Would love some help.

Thanks

UPDATE - Ok so I just noticed that my compute_capability is only 860.

[senayuksel]

In your codes, did you use any framework especially like tensorflow lite?

[MEATHEADBooYA]

I was speaking with some people and they claimed that Yolov7 is not as good as Yolov4. I think I will stick with Yolov4 myself.

[chaimacham]

please how can i improve my fps my yolov7 model when i run it with my custom dataset show 4fps and it is really low what can i do ?

[mizantednioc]

Are you running your test using an IP Camera or an mp4? yolov7 on my machine is fast only with my webcam

[rahmadyd]

can u show how to display fps on screen? thx

[NevaEdizUfuk]

we need custom object detection on python 3.9 and last version tensorflow please you haven't like video !

[humanized]

Can u try demonstrating a custom dataset training in kaggle? Kaggle is better than google colab as it gives u 30 gpu hours per week and doesn’t restart like google colab
Also can u show us how to implent Yolov7-tiny and a python code demo?

[pearcexx]

How can you demonstrate the FPS values ?