Image Transmission Main-Object Selection for Precise Video Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object recognition technologies apply uniform criteria across multiple videos, leading to a vast number of recognized objects, making it difficult for users to monitor or perform detailed searches for objects satisfying specific conditions, and existing best shot technologies do not consider back-end device requirements, potentially deteriorating camera performance due to calculation burdens.
Innovation Solution
An image transmission device and method that dynamically selects a main object based on metadata scoring, resource availability, and back-end requests, transmitting metadata and image frames or best shots tailored to the recipient's needs, using AI or motion sensing for object recognition and adjusting criteria based on resource conditions and user inputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If uniform object recognition criteria are applied across multiple videos, then comprehensive object detection is achieved, but the number of recognized objects becomes vast making monitoring and detailed search difficult
Solution Approach 1:
The patent extracts only the most representative objects (main objects) from the vast set of recognized objects by calculating scores based on multiple metadata attributes and comparing them against reference values. This extraction process filters out less important objects while preserving comprehensive detection capability through metadata generation for all objects.
Solution Approach 2:
The patent applies different quality levels to different objects by generating detailed metadata for all detected objects but only selecting and transmitting information for main objects that meet specific criteria. This allows high-quality detailed information for important objects while reducing data volume for less important ones.
2Ease of operation
If best shot technology generates partial images in the camera device, then representative images are provided, but calculation burden deteriorates camera performance when specifications are insufficient
Solution Approach 1:
The patent introduces a score calculation mechanism as an intermediary between object detection and best shot generation. By calculating comprehensive scores based on multiple metadata attributes and comparing against reference values, the system objectively determines main objects without requiring complex image processing in the camera device, thus reducing computational burden.
Solution Approach 2:
The patent performs preliminary object detection and metadata generation for all objects, then uses score calculation to pre-identify main objects before generating best shots. This preliminary scoring action reduces the complexity of subsequent image processing by already having identified which objects require detailed representation.
3Loss of information
If all recognized objects are transmitted with full metadata, then complete information is provided to back-end devices, but data transmission volume and processing load increase
Solution Approach 1:
The patent applies partial action by generating complete metadata for all detected objects but only selecting and transmitting detailed information for main objects that exceed reference score values. This allows the system to maintain complete information availability in metadata while reducing actual transmission volume by selectively conveying only the most important object data.
4Measurement precision
If dynamic selection of main objects based on multiple criteria is implemented, then search accuracy is improved, but calculation complexity increases
Solution Approach 1:
The patent segments the object selection process into distinct stages: metadata generation for all objects, score calculation using multiple attributes, reference value comparison, and main object selection. This segmentation allows each stage to be processed independently and efficiently, reducing overall computational complexity while maintaining high search accuracy through multi-criteria evaluation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An image transmission method performed by instructions in an image transmission device including a processor and a memory storing the instructions executable by the processor includes, receiving a video, recognizing an object from the received video, generating metadata of the recognized object, calculating a score of the recognized object using the generated metadata; selecting the recognized object as a main object when the score exceeds a reference value, transmitting the metadata on the selected main object, and transmitting one of an image frame including the main object in the video and a best shot cropped from the image frame.