Correct mutation adding the missing parameters

[iker-lluvia]

Correct mutation considering the higher number of segmentation parameters compared to object detection.

Fixes #9730

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Updated hyperparameters mutation logic in YOLOv5 training algorithm.

📊 Key Changes

• Changed the index offset from 7 to 12 for accessing hyperparameters values during mutation.
• Adjusted the keys sent to print_mutation from using all keys to a specific subset (keys 4 to 16).

🎯 Purpose & Impact

• These changes likely aim to refine the genetic algorithm that mutates hyperparameters for better training performance. 🧬
• By altering how hyperparameters are mutated, there could be an impact on the efficiency and effectiveness of the model training process. 🏋️‍♂️
• Users may experience improved model accuracy or training times due to these optimizations. 🔍🕒

[glenn-jocher]

@iker-lluvia thanks for the PR. You should not redefine dicts or lists that already exist in YOLOv5. Duplicating code increases maintenance and needlessly grows the codebase. The segmentation keys are here:

yolov5/utils/segment/metrics.py

Lines 180 to 200 in 85f6019

	KEYS = [
	'train/box_loss',
	'train/seg_loss', # train loss
	'train/obj_loss',
	'train/cls_loss',
	'metrics/precision(B)',
	'metrics/recall(B)',
	'metrics/mAP_0.5(B)',
	'metrics/mAP_0.5:0.95(B)', # metrics
	'metrics/precision(M)',
	'metrics/recall(M)',
	'metrics/mAP_0.5(M)',
	'metrics/mAP_0.5:0.95(M)', # metrics
	'val/box_loss',
	'val/seg_loss', # val loss
	'val/obj_loss',
	'val/cls_loss',
	'x/lr0',
	'x/lr1',
	'x/lr2',]

[iker-lluvia]

@iker-lluvia thanks for the PR. You should not redefine dicts or lists that already exist in YOLOv5. Duplicating code increases maintenance and needlessly grows the codebase. The segmentation keys are here:

yolov5/utils/segment/metrics.py

Lines 180 to 200 in 85f6019

	KEYS = [
	'train/box_loss',
	'train/seg_loss', # train loss
	'train/obj_loss',
	'train/cls_loss',
	'metrics/precision(B)',
	'metrics/recall(B)',
	'metrics/mAP_0.5(B)',
	'metrics/mAP_0.5:0.95(B)', # metrics
	'metrics/precision(M)',
	'metrics/recall(M)',
	'metrics/mAP_0.5(M)',
	'metrics/mAP_0.5:0.95(M)', # metrics
	'val/box_loss',
	'val/seg_loss', # val loss
	'val/obj_loss',
	'val/cls_loss',
	'x/lr0',
	'x/lr1',
	'x/lr2',]

Sorry, I wanted to keep the changes to a minimum and the result to be similar to how the general train.py, the one that doesn't correspond to segmentation, is programmed.
I will try to avoid duplications and create a new commit.

[glenn-jocher]

@iker-lluvia PR looks good, merged!