Pinned Repositories
acrobot
With a 9-DoF IMU (inertial measurement sensor), M2 microcontroller, motors and wheels, I made a self-contained balancing robot that only touch the ground through two coaxially-mounted wheels.
Fast_ICP
Interactive Closest Point for 3D reconstruction, support normal space sampling, closest point matching, point2point/point2plane error minimization
gmm_color_object_detection
a hybrid Gaussian Mixture Model for color segmentation, and connected component analysis for object detection
hmm_gesture_recognition
a left-to-right Hidden Markov Model for cellphone gesture recognition, using IMU data
particle_filter_SLAM
SLAM with occupancy grid and particle filter, using lidar, joints, IMU and odometry data from THOR humanoid robot
resnet_food_recognition
This project implements ResNet50 in keras and applies transfer learning from Imagenet to recognize food. It achieves 77.25% Top1 and 92.90% Top5 testing accuracy after 9 training epochs which takes only 5 hour.
structure_from_motion
Structure from Motion Pipeline to estimate camera motion and reconstruct 3D objects
ukf_orientation_estimation
a quaternion-based Unscented Kalman Filter on IMU to estimate quadrotor orientation. With estimates and camera data, a sphere panorama is generated by image stitching
vision_learning
Deep Learning for Computer Vision
visual_odometry
visual odometry on KITTI dataset, Monocular 2D-2D and Stereo 2D-3D implemented
MaxLing's Repositories
MaxLing/particle_filter_SLAM
SLAM with occupancy grid and particle filter, using lidar, joints, IMU and odometry data from THOR humanoid robot
MaxLing/visual_odometry
visual odometry on KITTI dataset, Monocular 2D-2D and Stereo 2D-3D implemented
MaxLing/ukf_orientation_estimation
a quaternion-based Unscented Kalman Filter on IMU to estimate quadrotor orientation. With estimates and camera data, a sphere panorama is generated by image stitching
MaxLing/Fast_ICP
Interactive Closest Point for 3D reconstruction, support normal space sampling, closest point matching, point2point/point2plane error minimization
MaxLing/structure_from_motion
Structure from Motion Pipeline to estimate camera motion and reconstruct 3D objects
MaxLing/gmm_color_object_detection
a hybrid Gaussian Mixture Model for color segmentation, and connected component analysis for object detection
MaxLing/hmm_gesture_recognition
a left-to-right Hidden Markov Model for cellphone gesture recognition, using IMU data
MaxLing/resnet_food_recognition
This project implements ResNet50 in keras and applies transfer learning from Imagenet to recognize food. It achieves 77.25% Top1 and 92.90% Top5 testing accuracy after 9 training epochs which takes only 5 hour.
MaxLing/acrobot
With a 9-DoF IMU (inertial measurement sensor), M2 microcontroller, motors and wheels, I made a self-contained balancing robot that only touch the ground through two coaxially-mounted wheels.
MaxLing/reinforcement_learning
MaxLing/vision_learning
Deep Learning for Computer Vision
MaxLing/hockey_robot
From mechanical design, circuit design to programming, my team constructed 3 fully autonomous robots to play hockey in Penn robockey cup 2016.
MaxLing/orchestra
I designed and constructed a battery-powered, GF1004 speaker-based instrument capable of playing a tone commanded by wireless signal.
MaxLing/pix2pix_refocus
MaxLing/test_openMVG-MVS
MaxLing/tracking_by_colorizing
https://arxiv.org/abs/1806.09594