18169281. IMAGE PROCESSING APPARATUS, IMAGE PROCESSING METHOD, AND IMAGE PROCESSING COMPUTER PROGRAM PRODUCT simplified abstract (KABUSHIKI KAISHA TOSHIBA)
Contents
- 1 IMAGE PROCESSING APPARATUS, IMAGE PROCESSING METHOD, AND IMAGE PROCESSING COMPUTER PROGRAM PRODUCT
- 1.1 Organization Name
- 1.2 Inventor(s)
- 1.3 IMAGE PROCESSING APPARATUS, IMAGE PROCESSING METHOD, AND IMAGE PROCESSING COMPUTER PROGRAM PRODUCT - A simplified explanation of the abstract
- 1.4 Simplified Explanation
- 1.5 Potential Applications
- 1.6 Problems Solved
- 1.7 Benefits
- 1.8 Potential Commercial Applications
- 1.9 Possible Prior Art
- 1.10 Unanswered Questions
- 1.11 Original Abstract Submitted
IMAGE PROCESSING APPARATUS, IMAGE PROCESSING METHOD, AND IMAGE PROCESSING COMPUTER PROGRAM PRODUCT
Organization Name
Inventor(s)
Tomoyuki Shibata of Kawasaki (JP)
IMAGE PROCESSING APPARATUS, IMAGE PROCESSING METHOD, AND IMAGE PROCESSING COMPUTER PROGRAM PRODUCT - A simplified explanation of the abstract
This abstract first appeared for US patent application 18169281 titled 'IMAGE PROCESSING APPARATUS, IMAGE PROCESSING METHOD, AND IMAGE PROCESSING COMPUTER PROGRAM PRODUCT
Simplified Explanation
The patent application describes an image processing apparatus that includes acquisition, pseudo label estimation, and learning units to identify attributes of unlabeled training data.
- The acquisition unit acquires unlabeled training data with images lacking correct attribute labels.
- The pseudo label estimation unit estimates pseudo-labels for the attributes of the images based on an identification target region.
- The learning unit learns a model to identify the attributes using labeled training data with assigned pseudo-labels.
Potential Applications
This technology could be applied in various fields such as image recognition, object detection, and pattern recognition.
Problems Solved
This technology addresses the challenge of training models with unlabeled data by estimating pseudo-labels for attributes, improving the learning process.
Benefits
The use of pseudo-labels for unlabeled data can enhance the accuracy and efficiency of attribute identification in images, leading to better performance of machine learning models.
Potential Commercial Applications
Potential commercial applications of this technology include automated image tagging, content-based image retrieval, and quality control in manufacturing processes.
Possible Prior Art
Prior art in the field of semi-supervised learning and active learning methods for training machine learning models with unlabeled data may be relevant to this technology.
Unanswered Questions
1. How does the accuracy of attribute identification compare when using pseudo-labels versus manually labeled data? 2. What are the computational requirements for implementing this image processing apparatus in real-time applications?
Original Abstract Submitted
According to one embodiment, an image processing apparatus includes one or more hardware processors configured to function as an acquisition unit A, a pseudo label estimation unit B, and a learning unit C. The acquisition unit A acquires unlabeled training data including an image to which a correct label of an attribute is unassigned. The pseudo-label estimation unit B estimates a pseudo-label, which is an estimation result of the attribute of the image of the unlabeled training data, based on an identification target region according to a type of the attribute to be identified by a first learning model to be learned in the image of the unlabeled training data. The learning unit C learns the first learning model identifying the attribute of the image by using first labeled training data with the pseudo-label being assigned to the image of the unlabeled training data.
