Object analysis
Patent Information
- Application Number
- ES2020709290T
- Authority / Receiving Office
- ES · ES
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2019-02-13
- Filing Date
- 2020-02-11
- Publication Date
- 2026-07-07
- Estimated Expiration
- 2040-02-11
AI Technical Summary
Existing object analysis methods in images and videos struggle with efficiency and accuracy, particularly in handling multiple resolutions and compression levels, and there is a need for improved real-time object detection and recognition systems.
A hierarchical system is employed, utilizing a tiered hierarchy of representations with different levels of quality (LoQs) for encoding and decoding signals, where object analysis is performed at multiple LoQs, leveraging convolutional neural networks (CNNs) and other neural networks to enhance detection and recognition accuracy and efficiency.
The hierarchical system enables more efficient and accurate object detection and recognition by optimizing processing time and resource use, allowing for real-time performance with improved accuracy by starting analysis at lower LoQs and refining at higher LoQs as needed.