Official Yolov7 Paper Explanation and Inference - Real-Time Object Detection At Its Zenith

[@LearnOpenCV]

Get expert guidance, insider tips n tricks and Create stunning images, learn to fine tune diffusion models, advanced Image Editing techniques like In-Painting, Instruct Pix2Pix and many more.
Join our Kickstarter campaign now! bit.ly/3JYh7A6

[@rodmallen9041]

Yes please, work on a video explaining in detail the different versions. And as usual, tks a lot for all your contributions to knowledge sharing on computer vision. :)

[@mishaalnaeem6135]

Thanks for the video. An in depth video or a comparison video would be great. 🙌

[@prakashchandra90]

We are working on comprasion video for different Yolo model. We will also create video for in depth explanation of Yolov7. Thnaks

[@kvnptl4400]

😊Thanks for the explanation of key modifications on Yolov7. Yes, we would like to have a video on comparing all YOLO models. 🌟

[@fathinazzaim6861]

dear sir, how did you get high fps with cpu?

[@sovitrath4735]

Hi. The tiny model can run at almost 15-17 FPS on a decent current generation CPU. Larger models need GPU to get real-time FPS.

[@leonallen4688]

thanks

[@iyshwaryakannan6677]

Hi sir, how to visualize the results of YOLOv7 using tensorboard as in YOLOv5

[@Thori45]

Can some tell my what the difference between the "normal models (tiny, normal, X)" and this "models with a 6 behind (W6, E6, D6)" is??

[@sovitrath4735]

The normal models are P5 models which are generally smaller compared to the W6, E6, and D6 models (also called as P6 models).

[@Thori45]

@@sovitrath4735 Ok thanks. So the P6 models just use a additional backbone layer for the anchor boxes right?

[@sovitrath4735]

@@Thori45 They surely have additional layers. But sometimes it is not as simple as adding just one layer. Maybe some other type of connections, like the auxliary connections used in YOLOv7 for the larger models.

[@Thori45]

@@sovitrath4735 Yeah I know, I'm writing my master thesis about a object detection topic but I'm not an expert so I haven't had a lot of experience and these papers (Yolov4 and Yolov7) are pretty hard to read because they mention many topics and references that many papers. To truly fully understand all of it is near impossible if you have not many years of experience this field I think 🤣

[@mohammadyahya78]

Thank you! Can I please fix YOLO as it detects drivers as pedestrians. Any solution to that? I don't think training will help as there will be always pedestrians and drivers.

[@thommone7172]

thanks for you work, i would like to know your opinion on "Yolov7 vs Yolov5" if possible. Based on my experimentation (object detection over videos), Yolov7 have better results than Yolov5 (precision + FPS). But Yolov5 is well maintained, have a lot of updates, a real roadmap for futur implementation, as well an enterprise backing it up. So i have a hard time deciding on wich solution i'm going to work with (long term).

[@iyshwaryakannan6677]

Please compare all the versions sir, its so useful

[@LearnOpenCV]

Hi, we are planning a comparison video
between all the YOLO models. Thank you for the suggestion!

[@LearnOpenCV]

📚 LINK TO BLOGPOST: learnopencv.com/yolov7-object-detection-paper-explanation-and-inference/
▶ LINK TO YOLO MASTERCLASS PLAYLIST: th-cam.com/play/PLfYPZalDvZDLALsG9o-cjwNelh-oW9Xc4.html

[@afriquemodel2375]

it doesnot work,

[@prakashchandra90]

Try using it on colab. If it does not work let me know.

[@afriquemodel2375]

Traceback (most recent call last):
File "C:\Users\aubin\Desktop\yolov7\train.py", line 24, in
import test # import test.py to get mAP after each epoch
File "C:\Users\aubin\Desktop\yolov7\test.py", line 12, in
from models.experimental import attempt_load
File "C:\Users\aubin\Desktop\yolov7\models\experimental.py", line 6, in
from models.common import Conv, DWConv
File "C:\Users\aubin\Desktop\yolov7\models\common.py", line 11, in
from torchvision.ops import DeformConv2d
File "C:\Users\aubin\Anaconda3\envs\yolov3\lib\site-packages\torchvision\__init__.py", line 7, in
from torchvision import datasets
File "C:\Users\aubin\Anaconda3\envs\yolov3\lib\site-packages\torchvision\datasets\__init__.py", line 1, in
from .lsun import LSUN, LSUNClass
File "C:\Users\aubin\Anaconda3\envs\yolov3\lib\site-packages\torchvision\datasets\lsun.py", line 2, in
from PIL import Image
ModuleNotFoundError: No module named 'PIL'

[@prakashchandra90]

Install Pillow to fix the error. Use the following command to install:
pip install Pillow

[@afriquemodel2375]

@@prakashchandra90 o work in jupyter notebook. I have same issue when i install pillow

[@afriquemodel2375]

@@prakashchandra90 is there any tutorial how to buil custom datasets ??

[@prakashchandra90]

@@afriquemodel2375 We will write a article on data annotation for object detection soon. Please subscribe the channel to get notified.