Official YOLO v7 Custom Object Detection Tutorial | Windows & Linux

This is a complete YOLO v7 custom object detection tutorial, starting from annotating the custom dataset, setting up environment for training custom model, and any modifications required in the official repository files. The method is generic enough to train all seven variations of official YOLO v7 models.

** Code and custom dataset is available for our Patreon Supporters**
---------------------------------------------
► Time Stamps:
Introduction: (0:00)
Creative environment: (0:30)
Develop custom dataset: (1:00)
Annotating custom dataset in YOLO format: (2:47)
Setup YOLO v7 repository: (4:40)
Modify YOLO v7 official files: (6:42)
Train YOLO v7 custom: (8:42)
Run custom object detector on images: (10:44)
Run custom object detector on videos: (12:08)
---------------------------------------------
► Links:
---------------------------------------------
Want to discuss more?

#TheCodingBug
---------------------------------------------
► My Other Tutorials:
---------------------------------------------
---------------------------------------------
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!

[connectrRomania]

Man keep these simple efficient videos about SOTA CV models, thanks alot ❤

[amirhosinipur]

Thanks a lot for a very effective, fast and complete, tutorial, not wasting people time.

[vgtfcfm]

what a nice tutorial! This video makes me surprised and more easier to realize the mechanism of yolov7.

[rahmadyd]

Thank you, dude. I have already completed my thesis using the algorithm. God Bless U

[sicco]

Have not tried it yet but its the best tutorial I could find. Very good content!

[Davida_dev]

Thank you! Helped me a lot in my school project :D Amazing video :D

[sanawar]

Thanks a lot. I have been trying to train a model for a week and finally I found this gem. I was having an issue regarding numpy: AttributeError: module 'numpy' has no attribute 'int'. The problem was with the latest version of numpy. I ran "pip install --upgrade numpy==1.23.5" and it fixed the issue.

[sootsyr]

Thank you very much!Everything works perfectly (except for a few issues here and there, but it was all solved at the end), i really appreciate this!

[hughmungus]

Amazing my friend thank you for posting this

[HasanAYousef]

great one appreciate your efforts, thanks a lot

[angelospapadopoulos]

hello great video. Question: how the official YOLOv7 wants the bounding box format to be? Just like COCO format [xmin, ymin, width, height] or like previous YOLOs [xcenter, ycenter, width, height]?

[brunospfc]

LabelImg only create boxes, it is possible to do Instance segmentation in this Yolo ? What label-software to use?

[journeywithwolf]

I followed the steps that you mentioned in the video and I tried 4-5 times to train the model but in the end, after training it did not detect Jack Sparrow's face in the image when trying to give an input image

[behnammashhadi]

Thank you so much, You solve my problem with your video. I wish you could do also real-time detection with camera :(

[nurulnajwakhamis]

hello sir. thank you for the explanation. i try to deploy with my 30 minutes video duration but turns out the detection video becomes 16 minutes. does it related to fps?

[Vislooo]

I tried training for tomatoes and tomato leafs but all the bounding boxes came up labeled as tomatoes. Is there a way to make it so leafs are detected as leafs and tomatoes are detected as tomatoes? I labeled them differently in labelImg so i dont know how they came up with same label after running detection. @TheCodingBug

[ranati]

Hi,
I am having this error
RuntimeError: CUDA error: no kernel image is available for execution on the device
Whereas my cuda and pytorch versions are the same
Name: torch
Version: 1.11.0+cu113
I had 11.6 but i changed it to 11.3 now it should work but its not and iam stuck now.

EDIT: I have gtx 770 which is old. But i want to run yolo can i run only on cpu ? As i have installed only cpu pytorch but it gave this error:
AssertionError: CUDA unavailable, invalid device -1 requested

Please Assist ASAP

[mihirshah]

Hey, can you tell what is the format for annotation which yolov7 accepts, I cannot use labelmg as it is company work. I saw on google the format it accepts is class, x_centre, y_centre, w, h all normalized. Even after using this format I am not getting any detections. Can you help me by telling the format which yolov7 accepts.

Edit - I figured my annotation format is correct, but still I am not able to get even a single detection on any image. what do you think might be the reason??

[pierfrancescomariasanti]

Hi Sir, as told you in the recent past, thank you so much for your effort producing this awesome minicourses. I'd like so much if you could produce a Yolov7 or Yolov8 detect video for Raspberry Pi. What's your thought regarding this idea?

[khenpahilanga]

Hello, thanks for this video, straight to the point and no time wasted, I just wanna ask after the first training of custom dataset. how do i add more dataset unto the first trained custom dataset yolo? do i use the best weights if i wanna continue adding more dataset to improve accuracy and more classes? becuz i wanna train my yolo while it retains the previous data is was trained with without starting from scratch, so i was wondering if u could help me with that. Thank you alot :)