How To Prepare Datasets For Training YOLOv5 Object Detection- Official - YOLOV5 Training

Watch this video to learn how to prepare a dataset for training a Custom Object Detector using the official YOLOv5.
We discuss the dataset format for YOLOv5 as well as provide tips on how you should prepare a dataset that allows your model to learn efficiently. We also talk about how to perform a train-val-test split and discuss how variance in data is a primary ingredient for learning a robust model.

❓ FAQ
How would you prepare a dataset for object detection?
How do I train my YOLOv5 with a custom dataset?
What is the name of the dataset YOLOv5 model used to detect objects?
How many epochs should I train YOLOv5?

📚 Link to the blog post of this video:

🎵 YOLO Master Class Playlist:

⭐️ Time Stamps ⭐️
0:00-00:20: Introduction
00:20-2:38: How to prepare your data
2:38-6:12: How to create a Dataset
6:12-8:18: Directory Structure
8:18-9:02: Tools required

🤖 Learn from the experts on AI: Computer Vision and AI Courses
YOU have an opportunity to join the over 5300+ (and counting) researchers, engineers, and students that have benefited from these courses and take your knowledge of computer vision, AI, and deep learning to the next level.🤖

#️⃣ Social Media #️⃣

🔖Hashtags🔖
#AI #Yolov5 #machinelearning #objectdetection #deeplearning #computervision #Dataset #DataPreparation #DeepLearningTips #TrainValSplit #DataUnderstanding

[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.

[fzgarcia]

Thank you Satya for clear information, your way to teach is the difference! Top!

[ameliebe]

I am watching the whole YOLOv5 series to learn how to customize my object detection model and I'm finding it incredibly useful. Thank you so much for the high quality content you share with us. Greetings from México!

[pyaephyo]

Thank you so much. I really like your videos.

[perriscalderon]

THANKS for this video, can you tell me how you get around the "catastrophic forgetting" notice when adding new classes and training data to an existing model?

[Rikimkigsck]

This is the second video of yours I have watched and you explain everything so beautifully. I am a complete beginner want to use a subset of Google's Open Images but can't figure out how to create YOLO format annotations.

[jubinthomas]

Hello sir, what exactly does 1 file per image mentioned @ 01:20 means

[LearnOpenCV]

Chapters
0:00-00:20: Introduction
00:20-2:38: How to prepare your data
2:38-6:12: How to create a Dataset
6:12-8:18: Directory Structure
8:18-9:02: Tools required

[fzgarcia]

Satya, can you help me with one question? What is the best pratice when using zoom level in the step of mark the object, drawing the bounding boxes? For example I'm labeling a class that appears in front of the image and there's another class that is in the end (3d) of the image and for it I need to use the zoom. Sometimes I made this questions, until which level I can use the zoom because pixels will be explosed and than maybe It's not good to use a high level of zoom. Thanks.

[harshawardhankulkarni]

I have prepared data set in XML format can be used for yolov5 algorithm

[zy.r.]

Hi! Thanks for the video. I trained custom object. But when two objects come too close it detecting as 1 object. How can I fix this?

[rihababbessy]

is there any script can do the same work so we don't waste time on doining this manually ?

[MIKO-pebu]

Is the picture of images matter? for example if i want training using img 640 can the data larger or smaller than training size?

[diego-itechsegurancaeletro]

Very nice and well done explanation.. Congrats.. Do you do Consulting or custom trains ? How can I contact you? Thanks

[abderrahmenmelliti]

Thank you so much, I just have a question, is that okay for txt file that has the information of annotation has the shape of a polygon not a rectangle?

[YigalBZ]

Do I really have to mark every picture with the object myself? That is a lot of non-automated work. Can I simply place pictures which include only the object and let the software do the labeling?

[DaupiejoKuowohko]

Great content (much appreciated) but very annoying background music.