17455514. DATA AUGMENTATION FOR MACHINE LEARNING simplified abstract (International Business Machines Corporation)

From WikiPatents
Jump to navigation Jump to search

DATA AUGMENTATION FOR MACHINE LEARNING

Organization Name

International Business Machines Corporation

Inventor(s)

Hiromi Kobayashi of Tokyo (JP)

Masaharu Sakamoto of Tokyo (JP)

Aya Nakashima of Tokyo (JP)

Kazuya Hirayu of Tokyo (JP)

Sho Ikawa of Kanagawa-Ken (JP)

DATA AUGMENTATION FOR MACHINE LEARNING - A simplified explanation of the abstract

This abstract first appeared for US patent application 17455514 titled 'DATA AUGMENTATION FOR MACHINE LEARNING

Simplified Explanation

Abstract: This patent application describes a computer-based method for training an image recognition model. The method involves using a set of processor units to create a saliency map of an original image. The saliency map is then superimposed on the original image to create an augmented image, which is used to train the image recognition model.

  • The method involves using processor units to create a saliency map of an original image.
  • The saliency map is a visual representation that highlights the most important regions or features of the image.
  • The saliency map is then superimposed on the original image, creating an augmented image.
  • The augmented image is used to train an image recognition model, improving its ability to identify and classify objects in images.

Potential Applications:

  • This technology can be applied in various fields that require image recognition, such as autonomous vehicles, surveillance systems, and medical imaging.
  • It can enhance the accuracy and efficiency of image recognition systems used in security and monitoring applications.
  • The method can be used to improve the performance of image recognition models in industries like e-commerce, where accurate object recognition is crucial for product identification and recommendation systems.

Problems Solved:

  • Traditional image recognition models may struggle to accurately identify objects in complex or cluttered images.
  • The saliency map generated by this method helps to highlight the most relevant regions of an image, improving the model's ability to focus on important features.
  • By superimposing the saliency map on the original image, the augmented image provides additional visual cues to the model, aiding in its training process.

Benefits:

  • The use of saliency maps and augmented images can significantly improve the accuracy and performance of image recognition models.
  • By training the model with augmented images, it becomes more robust and capable of accurately identifying objects in various scenarios.
  • This method allows for more efficient training of image recognition models, reducing the time and resources required for training.


Original Abstract Submitted

A computer implemented method trains an image recognition model. A set of processor units creates a saliency map of an original image. The set of processor units superimposes the saliency map on the original image to form an augmented image, wherein the augmented image is used to train the image recognition model.

Retrieved from "http://wikipatents.org/index.php?title=17455514._DATA_AUGMENTATION_FOR_MACHINE_LEARNING_simplified_abstract_(International_Business_Machines_Corporation)&oldid=23530"