CODEWITHSUNDEEP
validationtraining-datayolov8custom-training
Surya Giri
Surya Giri

Reputation: 1

Why is the YOLOv8 model not detecting objects in the training data but detecting them in the validation data instead?

Every time I tried to train a model it takes the images from the validation folder. Actually I have 299 images in training folder and 85 images in validation folder but during the training I found that 85 images is taken and you can see the instances of the images. . I searched everywhere and double checked the yaml file, looks everything is good but I am getting each time only 85 images during training.

I am expecting to run train folder with 299 images not the 85 images from the validation folder.

Upvotes: 0

Views: 266

Answers (1)

TK_pdbm
TK_pdbm

Reputation: 1

Assuming you use Ultralytics YOLOv8

The output of YOLO training cycle reports loss values and other metrics for validation set, as they are more representative of model's performance than training losses.

Actual training happens using training set, in your case on 299 training images, so your setup looks correct. If you check results.csv file you can actually see both training and validation losses.

Upvotes: 0

Related Questions