Dictionary Control Section Bitmap Tracking for Image Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image recognition systems face inefficiencies due to the need to read extensive dictionary data from external memory, leading to prolonged processing times and increased power consumption, as they store vast amounts of data for various object types and postures, making real-time recognition challenging.
Innovation Solution
The implementation of a dictionary control section within the image recognition apparatus that utilizes a bitmap to track referenced dictionary elements, allowing only necessary data to be read from external memory, reducing unnecessary data acquisition and enhancing processing efficiency by using a bus control finite state machine, dictionary data buffer, and bitmap to manage dictionary element access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If dictionary data is stored in external memory to reduce memory requirements, then memory cost is reduced, but data acquisition time increases
Solution Approach 1:
The dictionary data is segmented into multiple blocks, and a bitmap is created to indicate which blocks are needed for recognition. This allows the system to read only the necessary blocks from external memory rather than the entire dictionary, reducing both memory requirements and data acquisition time.
Solution Approach 2:
The bitmap indicating which dictionary blocks are needed is generated in advance before data acquisition. This preliminary action enables the system to efficiently read only the necessary dictionary blocks from external memory, significantly reducing data acquisition time while maintaining recognition accuracy.
2Reliability
If all dictionary data is read from external memory for recognition, then recognition accuracy is maintained, but processing efficiency decreases
Solution Approach 1:
Instead of reading the entire dictionary data, the system reads only the specific blocks indicated by the bitmap. This local quality approach ensures that recognition accuracy is maintained for the necessary data while significantly improving processing efficiency by avoiding unnecessary data transfer and processing.
Solution Approach 2:
The necessary dictionary blocks are extracted from the external memory based on the bitmap indication. This extraction approach allows the system to maintain recognition accuracy by reading only essential data while improving processing efficiency by eliminating unnecessary data acquisition operations.
3Reliability
If extensive dictionary data is stored and processed, then recognition accuracy is improved, but power consumption increases
Solution Approach 1:
Only the necessary dictionary blocks are extracted and read from external memory based on the bitmap, rather than loading the entire dictionary. This reduces the amount of data processed and thereby decreases power consumption while maintaining recognition accuracy for the required recognition tasks.
Solution Approach 2:
The system performs partial action by reading only the necessary dictionary blocks indicated by the bitmap, rather than processing the complete dictionary data. This partial approach maintains sufficient recognition accuracy while significantly reducing power consumption by avoiding unnecessary data processing operations.
Data Source
AI summary
An image recognition apparatus in an embodiment includes a feature-value calculating section configured to calculate a feature value in a region of interest segmented from an image, a likelihood calculating section configured to calculate likelihood of an object present in the region of interest referencing the feature value including a plurality of feature value elements and dictionary data including a plurality of dictionary elements, and a dictionary control section configured to acquire the dictionary element corresponding to the feature value element exceeding a set value.