Vehicle Appearance Annotation for Faster Camera Model Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous driving systems face challenges in efficiently training computer vision models for vehicle detection and accurately identifying vehicle appearances in real-time environments.
Innovation Solution
An electronic device and method that enhance the training of a computer vision model by obtaining user-input attributes for potential vehicle appearances, storing this information in a data set, and using it to improve the model's efficiency in detecting vehicles through a camera system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a computer vision model is trained using automatically collected image data without user input, then the training process is efficient and fast, but the accuracy of vehicle appearance detection is insufficient
Solution Approach 1:
The system performs preliminary automatic collection and organization of image data containing potential vehicle appearances, then prepares it for user verification. This preliminary processing reduces the overall time required while ensuring that only relevant and accurate data is used for training, thus balancing training efficiency with detection accuracy.
Solution Approach 2:
The system implements a feedback mechanism where user corrections and verifications of automatically detected vehicle appearances are collected and used to iteratively improve the computer vision model. This feedback loop enhances detection accuracy over time while maintaining efficient training through automated processes.
2Measurement precision
If user input is required for every potential vehicle appearance in training data, then the detection accuracy improves, but the ease of operation deteriorates
Solution Approach 1:
The system applies partial user intervention by requiring user verification only for images that contain potential vehicle appearances, while automatically processing and excluding images without vehicles. This selective approach maintains high detection accuracy while significantly reducing the operational burden on users.
Solution Approach 2:
The system performs self-service by automatically identifying potential vehicle appearances in images and preparing them for user verification. This automated pre-processing reduces the manual work required and makes the overall process easier to operate while maintaining high accuracy through targeted user input.
3Reliability
If all detected visual objects are processed through the computer vision model, then the reliability of vehicle detection is high, but the productivity of the system decreases
Solution Approach 1:
The system applies different processing qualities to different visual objects based on their characteristics. Images containing potential vehicle appearances undergo thorough computer vision model processing for high reliability, while images without vehicles are quickly filtered out. This localized quality approach maintains high detection reliability for relevant objects while improving overall system productivity.
Solution Approach 2:
The processing pipeline is segmented into multiple stages: initial visual object detection, potential vehicle appearance identification, and detailed computer vision model processing. This segmentation allows the system to quickly eliminate non-vehicle objects and apply comprehensive analysis only to relevant cases, thereby maintaining high reliability while improving processing speed.
Data Source
AI summary
According to various embodiments, an electronic device include a display, an input circuit, at least one memory and at least one processor configured to obtain a first image; display, in response to cropping an area comprising a visual object corresponding to a potential vehicle appearance from the first image, fields for inputting an attribute for the area, wherein, the fields include a first field for inputting a vehicle type as the attribute and a second field for inputting a positional relationship between a subject corresponding to the potential vehicle appearance and a camera obtained the first image as the attribute; obtain information about the attribute, by receiving a user input for each of the fields including the first field and the second field through the input circuit; store a second image configured of the area in a data set for training a computer vision model for vehicle detection.