YOLOv7 | Object Detection on Custom Dataset

YoloV7 is new framework which can perform various computer vision tasks like Object Detection, instance segmentation, keypoints detection .

This is the only framework support YOLOv4 + Instance Segmentation in single stage style.

It is based on detectron2. Detectron2 is A PyTorch-based modular object detection library by Facebook.
Yolov7 supports the training of yolov6.

[arnavthakur]

Worth watching ma'am.
#BestArtificialIntellegence channel I came across

[wetanortm]

Hello Aarohi, very nice video thanks for it, but i have a question : after training the model i tried to test it in the same way you showed but when the result window opend i saw that it took just a small part of the image and predicted different labels than the one that i give, it pridectied ( person, bike ...) with small accurcy, i think the model was pretrained with these labels but i do not understand way i got just very small part of the photo in the test window???

[syukrikoko]

Thank you Miss Aarohi for the great tutorial.

I hope you can make another video regarding the summary function of YOLOv1 until YOLOv7.
so, my friends (students) and I can understand more about all the families of YOLO...
😊😄

[angelospapadopoulos]

on where did you load the pretrained weights?

[KHakimin]

how can i save the output of inference on images in a folder to another folder?

[anselmekadio]

I'm trying to train Yolov7 on custom data. My dataset consists of .tiff images. The training ends with the error "AttributionError: Image not found".
Do you have a solution to this problem?

[akshayakki]

Mam, can you please tell me how to calculate the overall accuracy of the particular trained yolov7 model after testing an image in Google colab.

[ariouathanane]

Thank you very much, i do exactly the same steps as u but it predict (bus, car, motocycle ..etc ) not my own labels...

[vidhumathur]

How can we use variable resolution/ dimension images to train this model?

[baveshbalaji]

Hey, great tutorial. Just a very trivial question, how would train-annotations.json file look? Because I would like to create my own custom dataset and so far in the yolo series, all I do is create one text file for an image. But here, it seems slightly different as there is only JSON file for the entire training folder. It would be really helpful if you could shed some light on this doubt.

[devfordevs]

how we convert/export this model to tflite model for using in edge device

[farhanayasmin]

I have changed the ITERATION STEP into 10k in yaml file, but it doesn't stop after 10k. Any suggestion ma'am?

[freescholerships_.]

Mam Aarohi!
Your All videos are helpful.
Thank you 🌹

[adwaitkesharwani]

Great video!
Where can I deploy this model to get an API endpoint?

[mayanightstar]

I have absolutely no idea why Youtube recommended this to me, but keep up the good work lol. yolo.

[praisyallen]

Hi mam, Please provide video for bench mark dataset, i mean how to upload and train and test the bench mark dataset for object detection especially lowlight images

[gemshunt]

Hi! How to reparameterize a model and convert to onnx?

[fahimmahafuzruhad]

Ma'am how can we check the accuracy of our model?

[abderrahmenmelliti]

Thank you so much, you are the first one who did a tutorial about yolov7, that's really awesome,
I just have two questions:

How to deploy the model with a video or with a webcam for real-time running.
If I have the same object and I want to specify an ID for each one, do you have an idea how to do that?

and thank you so much.

[arturgoraus]

Hi,
Nice overview and explanation!
One question...
I have my own dataset that I've already trained on YoloV5 where all the image coordinates are in separate folder and files (each for every image). Can I use this structure with separate text files to train YoloV7?
Thank you :)