Diaz Angga
Always Learning!
Results-driven junior control systems engineer transitioning into a computer vision engineer role with 2 years of experience in the automotive industry. Seeking to use Python and AI-related libraries to solve accessibility-related social issues.
- Fast Lane Detection in CARLA using BiSeNet V2
● Built an improved model in inference speed from my previous lane detection model using UNet architecture in CARLA datasets. ● Utilized BiSeNet V2 as the backbone of architecture; the focal loss function for semantic segmentation loss and the discriminative loss for instance segmentation loss. ● Implemented in Python using the TensorFlow and Keras libraries ● Was able to achieve mean IoU at around 78% with an inference speed of 73 fps, as opposed to the previous implementation of 50% mean IoU with 15.8 fps of inference speed.
- - Sign Language Recognition using I3D in WLASL Dataset
● Built sign language recognition in TensorFlow and Keras utilizing I3D architecture with the inputs of RGB and optical flow. ● Trained the model with the pre-trained weights on Kinetics. This resulted in 65.7% of top-1 accuracy
Ambition
In the future
I would like to utilize my abilities and experience in the past such as Python, TensorFlow, ROS, etc, in a working environment to improve efficiencies such as programming related to AI technology and autonomous driving technology.
CBSテクノ3 years
株式会社CBSテクノ
-
Control System Engineer (Dispatch Employee of CBS Techno, Ltd.) at Subaru, Ltd., Ota July 2020 — February 2021 Vehicle Communication Controller Team, Electronic Controller Development Division
Bandung Institute of Technology (ITB)5 years
Electrical Power Engineering
-
GPA: 3.21 of 4.00 Thesis Title: Study of Regenerative Braking as A Substitute to Frictional Braking System
ieeexplore.ieee.org
ascelibrary.org
