狠狠撸

deep learning cnn tracking prediction rnn autonomous driving detection intention segmentation gan camera trajectory reinforcement learning modularity driving behavior end-to-end aggressive courteous abstraction primitive transfer learning database lidar depth selfish pose estimation lstm stereo unsupervised regression behavior classification pose fusion transform preference inter-city supervised zebra crossing crowd imitation learning social imitation learning. gnn obstacle. depth fusion slam pedestrian bev data augmentation completion mapping calibration inpainting adas keypoints projection 2d-to-3d vo vio monocular rpn 3d proposal point cloud encoder-decoder network mask gradient blending equirectangular spherical sparse depth sensor distortion surround view panorama omnidirectional fisheye domain adaptation sensor fusion fcn 3d cuboid bounding box gru lane fitting lane detection ego lane prediction. rgb-d decoder multiscale loss perception planning norm lane localization encoder vanishing point reinforcement learning. multi-task gan. registration joint training lmnet hdnet ipod birdnet rt3d voxelnet regnet radar region proposal ssd pixor depthcn second yolo3d feature cost volume anchoring rgb multi-modal frontal view late fusion data pipeline localization semantic machine learning simulation embodied ai 3d object detection monocular camera self driving annotation tools for training data boosting weather dust fog snow rain denoising coordconv sar focal loss fpn pointnet frustum pointnets doppler faster rcnn two stage rcs semi-supervised one stage pseudo label teacher-student constrastive learning active learning early fusion brightness constancy constraints local method large displacement field aperture problem focus of expansion horn schunck regularization and smoothness constraints global method descriptor matching small motion parametric model depth/disparity prediction image-based modeling 3-d object modeling view synthesis structure from motion and scene flow passive stereo vision disparity estimation graph cut sparse coding belief propagation census transform cost aggregation guided filter rectification disparity refinement single image optic flow estimation hamming distance dynamic programming semi-global matching fundamental matrix essential matrix optic flow lucas-kanade algorithm edge aware filters deepedge image denoising colorization. joint image processing image deconvolution dehazenet image/depth superesolution deepcontour image restoration holistically-nested edge detection boundary detection image enhancement modeling recognizing learning road network scenario traffic end-to-end. total variation route planners driving models e2e learning driving policies driving path failure prediction for autonomous driving lidar conditional affordance learning conditional imitation learning low level feature description and matching vehicle navigation trajectory features driver behaviors traffic constraints dynamic maneuvers autonomous planning autonovi image-based relocalization for slam denoiser prior artifact reduction cameras annotation sensor data behaviors simulating safety transformer conditional interactive hypotheses forecasting dynamics multi-task learning interaction graph data lake open source code behavior planning trajectory prediction closed loop data pipeline engine kodiac robotics aururo nuro ike pony tron waymo via tusimple inceptio telsa semi plus.ai big data dexerity generalization robto ai llm agent memory reflection nerf diffusion model world model llms foundation model cloud long tailed distribution data close loop conntrol decision making robotaxi adverse outliers dynamic scenario modeling scenarios clustering safety testing autonomous vehicles scenario-based self-supervised learning model testing and verification large scale model training platform data driven models cloud computing infrastructure and big data proces residually connected top-down module context semantic segmentation disparity pyramid stereo matching network dispsegnet segstereo object detection cornernet-lite keypoint triplets centernet objects as points ga-rpn region proposal by guided anchoring center and scale prediction foveabox fully convolutional one-stage fcos feature selective anchor-free fsaf extremenet cornernet yolo densebox unitbox single view metrology sfm pnp aspect graph commercial truck 3d objeect detection monocular image distance occupancy sim-to-real unprojecting 3d 2d siamese network dtection-and-tracking multiple object tracking single object tracking self-supervised continual learning. knowledge transfer uncertainty modeling and estimation instance curriculum learning 3-d object detection uncertainty modeling open world multiple head backbone dpm color interpolation guided upsampling selfish.