Skip to content
Snippets Groups Projects
Select Git revision
  • main default protected
  • castelldefels
  • Esplugues_demo
3 results
You can move around the graph by using the arrow keys.
Created with Raphaël 2.2.05Mar26Apr251895Sep21Jul6Apr16Mar16Feb17Jan1695421Dec2010Oct767Sep20Jul19141312828Jun13May422Apr131Mar3029252423222118171410714FebAdd gitignore to ignore build dirsmainmainFixed bugRenamed parameterAdded parameters for low obstacle detectionCorrected error-> it is not possible to save the dataset AND use the algorithm simultaneouslyChange code order to permit storing dataset AND using the network for segmentationCleaning unused codeMerge branch 'main' of https://gitlab.iri.upc.edu/idelpino/kf_based_terrain_analysis into mainAdd LICENSEChanges to make it compile in ubuntu 20.04castelldefelscastelldefelsModified the CMakeFile.txt to include the external repository for the rapidcsv library and install the necessary files.some work in representation and elimination of graph nodes that are classified as obstacles by the neural netfixed bug in overhanging obstacles and improved visualizationdebugging labelling in not analyzed areasadded neural network of different possible sizesmulticlass preditionsadded loading of neural net from filedumping dataset to file for nn training in matlabfeatures extracted!working in neural network integrationadded constants to represent gt labels in colors in the point cloudcreating ground references in median points instead of centroidsdebugged!more cleaningcleaning old functions and removing unused parameterscleaning unused codedocumenting some functionsadded field for mean intensityadded full covariance use in mahalanobis distance computation, however it is disabled because it slows the algorithm and do not improve resultsadded intensity mean and variance in elevation pointcloudfwork in progressimproved optimized elevation cloud generation to avoid running out of memoryoptimized elevation cloud generation!improved fast labelling, now achieving even better results than with the slow method!added fast labelling mode, to enable it just set the number_of_reference_used_for_labelling to zeroadded heuristic to label points outside the local area analysis as ground or obstaclesbug fixed in time supervision --> removed ifs that make the object to be destroyed after the if sentenceadded semantic kitti classesadded some color labels for low obstacles and ground accumulatedadded class labels for low and negative obstaclesEsplugues_demoEsplugues_demo