README Segmentation Usage fixes

[glenn-jocher]

Fixes per #10288

Signed-off-by: Glenn Jocher glenn.jocher@ultralytics.com

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Update training and validation commands in YOLOv5 documentation.

📊 Key Changes

• Adjusted command lines for single and multi-GPU segmentation training.
• Updated the dataset used for validation from ImageNet-1k to COCO.

🎯 Purpose & Impact

• 🎯 Command line simplification: The removal of the --model argument simplifies the training command, potentially reducing confusion when initiating training sessions.
• 🔍 Improved clarity: Changing the validation dataset info ensures users are accurately informed about the dataset they will be working with, leading to more relevant validation metrics.
• ✅ Enhanced usability: These updates in the README file make the repository more user-friendly, enabling both new and existing users to better understand how to train and validate their segmentation models with YOLOv5.