Image Labeling Overlay for In-Vehicle Semantic Data Capture
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The generation of labeling data for training artificial neural networks in autonomous vehicles is time-consuming due to the manual process of identifying objects in images, which is inefficient and separates the data recording from the semantic information generation.
Innovation Solution
A method using a processing device to segment objects in images, generate graphical markers, and receive user inputs to create labeling data directly, allowing for efficient generation of semantic content through augmented reality and voice or gesture recognition during vehicle operation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual labeling by operators is used to generate labeling data, then labeling accuracy is improved, but time consumption increases significantly
Solution Approach 1:
The system enables operators to perform labeling tasks while the vehicle is in operation, utilizing idle time during normal driving. The operator interacts with the system through voice commands or gestures without needing to stop the vehicle or dedicate separate time for data collection, thus converting operational time into productive labeling time.
Solution Approach 2:
The system automatically captures images and pre-processes them during vehicle operation, preparing the data in advance for later labeling. This preliminary data capture and preparation eliminates the need for separate data collection trips and reduces the time required for the actual labeling process.
2Measurement precision
If manual labeling process is used, then labeling quality is maintained, but productivity decreases
Solution Approach 1:
The system enables continuous data collection and labeling throughout vehicle operation rather than requiring separate dedicated time for these tasks. Labeling can occur continuously during normal driving, turning previously idle time into productive labeling time and significantly increasing overall productivity.
Solution Approach 2:
The vehicle serves multiple functions simultaneously: it performs its primary transportation function while also collecting image data and generating labeling data through operator interaction. This multi-functionality eliminates the need for separate data collection missions and increases productivity without compromising labeling quality.
3Measurement precision
If data recording and semantic information generation are separated, then data accuracy is ensured, but process complexity increases
Solution Approach 1:
The system combines data recording and semantic information generation into a single integrated process. Images are captured during vehicle operation and labeled in real-time through operator interaction, merging two previously separate processes into one unified workflow that reduces complexity while maintaining accuracy through the operator's direct input.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention is concerned with a method for generating labeling data (27) that describe an image content of images (16) depicting at least one scene (12), wherein in a processing device (18) image data (15) are received from a imaging (14) and a segmentation unit (19) detects at least one object (17) in the image data (15) and a graphical processing unit (20) generates a respective graphical object marker (29) that marks the at least one detected object (17) and a display control unit (21) displays an overlay (35) of the at least one scene (12) and the at least one object marker (29) and an input reception unit (22) receives a respective user input (36) for each object marker (29), wherein the respective user input (36) provides the image content of the image region (26) marked by the object marker (29).