Size-color sensitive object, how to train correctly?

[thanhminhmr]

For context, I'm trying to create an OCR for simple text captcha. Currently I can generate simple text captchas with any specified size and color.

But since characters in captchas are sensitive to color and size (decoy characters have lower contrast than normal character, etc.) I don't know what setting I should train with. So my question is:

• Do the data loader stretch the image to the specified size, or will it try to fit the image in order to keep the original image ratio? (Captchas come with different size, but the character shape is the same, so ideally I should try to keep the original ratio of the image)
• Do the data augmenter change the color/ratio of the image, and how do I prevent that?

Note: This is what I mean when I say stretch or fit the image:

[glenn-jocher]

@thanhminhmr 👋 Hello! Thanks for asking about image augmentation. YOLOv5 🚀 applies online imagespace and colorspace augmentations in the trainloader (but not the testloader) to present a new and unique augmented Mosaic (original image + 3 random images) each time an image is loaded for training. Images are never presented twice in the same way.

The hyperparameters used to define these augmentations are in your hyperparameter file (default data/hyp.scratch.yaml) defined when training:

python train.py --hyp hyp.scratch.yaml

yolov5/data/hyp.scratch.yaml

Lines 1 to 33 in 90b7895

	# Hyperparameters for COCO training from scratch
	# python train.py --batch 40 --cfg yolov5m.yaml --weights '' --data coco.yaml --img 640 --epochs 300
	# See tutorials for hyperparameter evolution https://github.com/ultralytics/yolov5#tutorials
	lr0: 0.01 # initial learning rate (SGD=1E-2, Adam=1E-3)
	lrf: 0.2 # final OneCycleLR learning rate (lr0 * lrf)
	momentum: 0.937 # SGD momentum/Adam beta1
	weight_decay: 0.0005 # optimizer weight decay 5e-4
	warmup_epochs: 3.0 # warmup epochs (fractions ok)
	warmup_momentum: 0.8 # warmup initial momentum
	warmup_bias_lr: 0.1 # warmup initial bias lr
	box: 0.05 # box loss gain
	cls: 0.5 # cls loss gain
	cls_pw: 1.0 # cls BCELoss positive_weight
	obj: 1.0 # obj loss gain (scale with pixels)
	obj_pw: 1.0 # obj BCELoss positive_weight
	iou_t: 0.20 # IoU training threshold
	anchor_t: 4.0 # anchor-multiple threshold
	# anchors: 3 # anchors per output layer (0 to ignore)
	fl_gamma: 0.0 # focal loss gamma (efficientDet default gamma=1.5)
	hsv_h: 0.015 # image HSV-Hue augmentation (fraction)
	hsv_s: 0.7 # image HSV-Saturation augmentation (fraction)
	hsv_v: 0.4 # image HSV-Value augmentation (fraction)
	degrees: 0.0 # image rotation (+/- deg)
	translate: 0.1 # image translation (+/- fraction)
	scale: 0.5 # image scale (+/- gain)
	shear: 0.0 # image shear (+/- deg)
	perspective: 0.0 # image perspective (+/- fraction), range 0-0.001
	flipud: 0.0 # image flip up-down (probability)
	fliplr: 0.5 # image flip left-right (probability)
	mosaic: 1.0 # image mosaic (probability)
	mixup: 0.0 # image mixup (probability)

You can view the effect of your augmentation policy in your train_batch*.jpg images once training starts. These images will be in your train logging directory, typically yolov5/runs/train/exp:

train_batch0.jpg shows train batch 0 mosaics and labels:

Good luck and let us know if you have any other questions!