Abstract
Labeling is a very costly and time consuming process that aims to generate datasets for training neural networks in several functionalities and projects. In the automotive field of driver monitoring it has a huge impact, where much of the budget is used for image labeling. This paper presents an algorithm that will be used for generating ground truth data for 2D eye location in infrared images of drivers. The algorithm is implemented with many detection restrictions, which makes it very accurate but not necessarily very constant. The resulting dataset shall not be modified by any human factor and will be used to train neural networks, which we expect to have a very good accuracy and a much better consistency for eye detection than the initial algorithm. This paper proves that we can automatically generate very good quality ground truth data for training neural networks, which is still an open topic in the automotive industry.
Highlights
For the problem addressed in this paper, which is eye location in infrared images of drivers, this marked data can be very difficult to obtain
It is not meant to be used in real time scenarios but in the future it may be run for demonstration reasons against the neural network, which will be trained using its output
The first selection step is the black percentages rule, which will only keep those patches that meet the requirements described in Figure 5. This rule removes more than 95% of the possible eye patches in almost every frame
Summary
Training neural networks requires large datasets with good accuracy in order to have a general and precise detection model The generation of these datasets is a very expensive and time consuming aspect of this process, with whole teams of people working in data marking. For the problem addressed in this paper, which is eye location in infrared images of drivers, this marked data can be very difficult to obtain Another big problem in the labeling process is that hours of marking eyes on images is a very strenuous task and it always leads to natural human errors. This usually leads to a double checking process for the labeling, which means additional time and effort and it does not guarantee the expected quality improvement
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.