- Seoul, Korea
- https://www.linkedin.com/in/wolfapple
Pinned Loading
-
CarND-Behavioral-Cloning-Project
CarND-Behavioral-Cloning-Project PublicBuilt and trained a convolutional neural network for end-to-end driving in a simulator, using TensorFlow and Keras. Used optimization techniques such as regularization and dropout to generalize the…
-
CarND-Semantic-Segmentation
CarND-Semantic-Segmentation PublicLabeled the pixels of a road in images using a Fully Convolutional Network (FCN).
Python 1
-
CarND-Vehicle-Detection
CarND-Vehicle-Detection PublicDetect and track vehicles using color and gradient features and a support vector machine classifier.
Jupyter Notebook 1
-
CarND-Advanced-Lane-Lines
CarND-Advanced-Lane-Lines PublicBuilt an advanced lane-finding algorithm using distortion correction, image rectification, color transforms, and gradient thresholding. Identified lane curvature and vehicle displacement. Overcame …
Jupyter Notebook
-
traffic-sign-recognition
traffic-sign-recognition PublicBuilt and trained a deep neural network to classify traffic signs, using PyTorch. The highlights of this solution would be data preprocessing, trained with heavily augmented data and using Spatial …
-
pytorch-cam
pytorch-cam PublicPyTorch implementation of class activation mapping
Jupyter Notebook 1
If the problem persists, check the GitHub status page or contact support.