Pinned Repositories
auto_position_recovery
If you can create a 3D structure using a 3D marker, you can verify various things such as the distance between two objects and the positional relationship. However, if the three-dimensional structure created before and the current environment are different over time, if there is even one non-moving reference object, the three-dimensional structure can be restructured by combining the past structure and the present structure. This program is a tool that calculates its functions.
convert_head_pos_and_gaze_roi
In order to verify the performance of face recognition algorithms of other companies in 2021, we are going to review the relationship between GT data and recorded images of face recognition that we have developed.
data_parsing_and_notify_with_timestamps
In this project, I tried to parse and report the streaming data made with ADTF (Automotive Data and Time-Triggered Framework) DAT, and to notify according to TimeStamp. Additionally, we are trying to support outputting an image whether a normal notification has been reached.
extract_eye_gaze
purpose - extract face -> extract eye -> extract pupil( or with iris) -> extract gaze for eye tracking
stereo_camera_calibration_extension
In 2018, I was responsible for the production line camera setup of stereo camera calibration, and to ensure the performance of stereo camera calibration, an opencv-based inspection tool was created and various verifications were performed. The purpose was to compare the performance of the two camera calibration algorithms by inputting images and coordinate points as input. (OpenCV vs. other companies) In 2021, Distortion Rational, Thin Prime Model interface were additionally updated, The 2022 Fisheye model interface was also updated. In general, normal calibration is sufficient to cover, but If you want a normal wide angle or more precision, you can use the Rational model, For ultra-wide-angle cameras, we recommend using the Fisheye model. Since the method of calculating the distortion index is different for each model, it must be used separately.
yeolip
Config files for my GitHub profile.
yeolip's Repositories
yeolip/stereo_camera_calibration_extension
In 2018, I was responsible for the production line camera setup of stereo camera calibration, and to ensure the performance of stereo camera calibration, an opencv-based inspection tool was created and various verifications were performed. The purpose was to compare the performance of the two camera calibration algorithms by inputting images and coordinate points as input. (OpenCV vs. other companies) In 2021, Distortion Rational, Thin Prime Model interface were additionally updated, The 2022 Fisheye model interface was also updated. In general, normal calibration is sufficient to cover, but If you want a normal wide angle or more precision, you can use the Rational model, For ultra-wide-angle cameras, we recommend using the Fisheye model. Since the method of calculating the distortion index is different for each model, it must be used separately.
yeolip/convert_head_pos_and_gaze_roi
In order to verify the performance of face recognition algorithms of other companies in 2021, we are going to review the relationship between GT data and recorded images of face recognition that we have developed.
yeolip/extract_eye_gaze
purpose - extract face -> extract eye -> extract pupil( or with iris) -> extract gaze for eye tracking
yeolip/auto_position_recovery
If you can create a 3D structure using a 3D marker, you can verify various things such as the distance between two objects and the positional relationship. However, if the three-dimensional structure created before and the current environment are different over time, if there is even one non-moving reference object, the three-dimensional structure can be restructured by combining the past structure and the present structure. This program is a tool that calculates its functions.
yeolip/data_parsing_and_notify_with_timestamps
In this project, I tried to parse and report the streaming data made with ADTF (Automotive Data and Time-Triggered Framework) DAT, and to notify according to TimeStamp. Additionally, we are trying to support outputting an image whether a normal notification has been reached.
yeolip/yeolip
Config files for my GitHub profile.