Vehicle Gesture Detection Using Recognition Areas and Motion Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle operation detection systems face challenges in accurately detecting user gestures to trigger opening or closing operations, leading to potential errors in vehicle door control.
Innovation Solution
A vehicle operation detection device and method that utilize a trained model from machine learning to identify the position and movement of a user's body part within a recognition area, calculating displacement vectors to determine if a specific gesture is made, thereby accurately triggering door operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a trained model from machine learning is used to identify body part position and movement, then gesture detection accuracy is improved, but device complexity increases
Solution Approach 1:
The system performs preliminary actions by pre-training a machine learning model offline to recognize body part positions and gestures. The trained model is then deployed in the vehicle system, allowing it to accurately detect gestures without requiring complex real-time processing during actual use. This separates the complex training phase from the simpler detection phase in the vehicle.
Solution Approach 2:
The trained machine learning model acts as an intermediary between the camera input and the gesture recognition system. Instead of directly analyzing raw camera images to detect gestures, the system uses the trained model to first identify body part positions and movements, which then feed into gesture determination logic. This intermediary layer simplifies the overall system architecture while improving accuracy.
2Measurement precision
If displacement vector calculation is performed based on multiple images captured at time intervals, then gesture determination accuracy is improved, but calculation load increases
Solution Approach 1:
The system extracts only the necessary information from multiple captured images - specifically, the displacement vectors of identified body parts - rather than processing all image data. By focusing calculation efforts only on the relevant body part movements detected by the trained model, the system reduces overall calculation load while maintaining gesture determination accuracy.
Solution Approach 2:
The system performs calculations at selective time intervals rather than continuously processing every captured image. By capturing images at predetermined time intervals and only performing displacement vector calculations on these sampled frames, the system achieves sufficient gesture detection accuracy while significantly reducing the calculation burden compared to continuous processing.
Data Source
AI summary
A vehicle operation detection device includes a storage unit configured to store a trained model obtained by machine learning using training data in which an image captured in advance and a body part used for a gesture of a user are associated with each other, an entry determination unit configured to determine whether, based on a position of the body part in the image obtained by inputting a newly captured image into the trained model, the body part enters a recognition area set in an imaging area of a camera, and a gesture determination unit configured to calculate a displacement vector of the body part based on images captured at a time interval after it is determined that the body part enters the recognition area, and determine, in accordance with whether a direction of the displacement vector is a direction corresponding to the gesture, whether the gesture is made.