#Baiduyun xlog update#

#Baiduyun xlog update#

NOTE 2 ( update Mar 12, 2019): Today I’m releasing the second part of my journey with Structure from Motion (SfM) pipeline and OpenGL visualization for Apolloscape ZPark Sample implemented on C++. It’s is still an ongoing work where I intend to implement Vidloc, Pose Graph Optimization and Structure from Motion pipelines for Apolloscape Dataset in the context of the localization task. NOTE: If you want to jump straight to the code here is the GitHub repo. Implement PoseNet architecture for monocular image pose prediction and visualize results. This article covers the very beginning of the journey and includes the reading and visualization of the Apolloscape dataset for localization task.

I pick the self-localization problem and create the whole pipeline to localize a car which is based only on camera images. Task 1: Vision-based fine-grained lane markings segmentation.

It’s crucial for a system to know the exact pose (location and orientation) of the agent to do visualization, navigation, prediction, and planning.īaidu released its massive self-driving Apolloscape dataset in March and now has a couple of ongoing challenges for ECCV 2018 Workshop: Vision-based Navigation for Autonomous Driving conference: Localization is an essential task for augmented reality, robotics, and self-driving car applications.