Object Recognition Model Using Prediction Information
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in maintaining high recognition accuracy when using sparse point cloud data for object recognition in measured spaces.
Innovation Solution
A model generation device and method that includes a recognition unit for generating object recognition information from measurement data, a prediction information generation unit that adds situation information to the recognition information, and a model generation unit that uses machine learning to generate a model that improves object recognition accuracy by inputting prediction information from multiple frames.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If object recognition is performed using sparse point cloud data obtained by measuring a space, then the measurement process can be simplified and faster, but the recognition accuracy is lowered
Solution Approach 1:
The patent applies preliminary action by pre-processing the point cloud data to generate prediction information before the actual recognition task. Specifically, the system generates prediction information that includes situation information (spatial context, object relationships) and temporal information (frame sequences) in advance, which then guides the recognition process and improves accuracy without requiring denser measurements
Solution Approach 2:
The patent introduces prediction information as an intermediary element between the sparse point cloud data and the final object recognition result. This prediction information acts as a mediator that enriches the sparse measurements with contextual understanding, allowing the recognition system to achieve high accuracy despite the sparsity of the original measurement data
2Measurement precision
If prediction information including situation information is added to object recognition information, then the object recognition accuracy is improved, but the data processing complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the data processing into distinct modular components: a recognition unit that processes point cloud data, a prediction information generation unit that creates prediction information, and a model generation unit that combines them. This segmentation allows each component to handle specific tasks independently, making the overall complex processing manageable and systematic
Solution Approach 2:
The prediction information generation unit performs preliminary processing to extract and structure situation information from the point cloud data before it is used in the final recognition model. This pre-processing organizes the complex data into meaningful prediction information structures, reducing the complexity of subsequent processing steps
Data Source
AI summary
A model generation device of the present disclosure includes a recognition unit that generates, from measurement data including a plurality of frames obtained by measuring a space by a sensor, object recognition information representing information of an object recognized for each of the frames; a prediction information generation unit that generates, for each of the frames, prediction information in which situation information representing the situation at the time of measuring the space is added to the object recognition information; and a model generation unit that generates a model that inputs thereto a plurality of units of the prediction information corresponding to the plurality of frames and outputs an object recognition result in the space, by machine learning using the input units of prediction information, the output object recognition result, and correct data of the object recognition result.