Object Recognition Reliability Adjustment via User Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Electronic devices face inaccuracies in object recognition, leading to unreliable image processing and editing, particularly when performing object recognition tasks using machine learning or deep learning algorithms.
Innovation Solution
The electronic device improves object recognition reliability by associating identified objects with reliability scores and incorporating user input to enhance confidence in object identification, utilizing a processor to manage image processing and communication with external devices for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If machine learning or deep learning algorithms are used for object recognition, then object recognition capability is improved, but recognition accuracy and reliability deteriorate due to errors in identification
Solution Approach 1:
The system implements feedback by acquiring user inputs (such as voice commands or text) that indicate objects present in the image. These user inputs are compared with the automatically recognized objects, and the reliability scores are adjusted based on the correspondence between user-indicated objects and recognized objects. This feedback loop improves recognition accuracy over time.
Solution Approach 2:
The system performs self-service by automatically acquiring multiple objects from the image, generating reliability scores for each, and adjusting these scores based on user inputs without requiring manual intervention for each object. The system autonomously manages the object recognition and reliability assessment process.
2Quantity of substance
If multiple objects are automatically acquired from the image, then object detection coverage is improved, but reliability of individual object identification deteriorates due to potential errors
Solution Approach 1:
The system changes the parameter of reliability scoring by assigning different reliability scores to different objects based on various factors. The reliability score for each object is adjusted upward or downward depending on the correspondence between user inputs and recognized objects, allowing differential confidence levels across multiple detected objects.
3Reliability
If user inputs are incorporated to improve recognition accuracy, then reliability is improved, but system complexity increases due to additional processing requirements
Solution Approach 1:
The system achieves multi-functionality by using a single processor to perform both automatic object recognition through machine learning algorithms and user input processing. The same processor acquires images, identifies objects, processes user inputs (voice or text), and adjusts reliability scores, eliminating the need for separate dedicated hardware components for each function.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
According to embodiments, an electronic device comprising a processor configured to receive an image including one or more objects, acquire a first one or more of the received one or more objects, acquire one or more reliability measures associated with the acquired one or more first objects, receive an input including information having one or more words, acquire one or more second objects corresponding to at least a part of the one or more words, when there is at least one object corresponding to the one or more second objects among the one or more first object, adjust at least one reliability measure corresponding to the at least one object among the acquired one or more reliability measures, and recognize the one or more first objects by using an image recognition scheme at least based on the one or more reliabilities including the adjusted at least one reliability.