FAST Corner Detector Creation ----------------------------- Copyright (c) Edward Rosten 2006 This program will allow you to create your own FAST feature detectors. If you just wish to use FAST, then visit: http://mi.eng.cam.ac.uk/~er258/work/fast.html to get FAST in a variety of different forms. Use this program if you wish to train FAST on your own datasets or if you want to try your own optimizations or if you wish to target a different language. Requirements ------------ This pogram depends on libcvd and gvars. These can be downloaded from here: http://mi.eng.cam.ac.uk/~er258/cvd/index.html or obtained through CVS: cvs -d :pserver:anoncvs@cvs.sv.gnu.org:/cvsroot/libcvd co gvars3 cvs -d :pserver:anoncvs@cvs.sv.gnu.org:/cvsroot/libcvd co libcvd Both of these projects have a number of optional dependencies. This program does NOT require that any of these are present. Building -------- Edit the Makefile, in order to set up the include path and library path to point to libcvd and gvars3. Then "make" should do the job. Running ------- The operation of the program is controlled by "corner_learn.cfg" The training video can come from a variety of sources. The default configuration trains from a video stored as a list of files. If the training is successful, then the code will be printed to the standard output, so don't forget to redirect it to a file. Training on 120 768x288 frames resuires about 400M of memory and takes a few minutes on an Opteron at 2.4GHz. Extending --------- If you wish to target a different language, then look at tree_visitor.cxx If you wish to use a different strategy for selecting which question to ask, then examine find_best_split() in corner_learn.cxx Citations --------- These are the bibtex entries for the relavent papers: @inproceedings{rosten_2005_tracking, title = "Fusing points and lines for high performance tracking.", author = "Edward Rosten and Tom Drummond", year = "2005", month = "October", pages = "1508--1511", volume = "2", booktitle = "IEEE International Conference on Computer Vision", notes = "Oral presentation", url = "http://mi.eng.cam.ac.uk/~er258/work/rosten_2005_tracking.pdf" } @inproceedings{rosten_2006_machine, title = "Machine learning for high-speed corner detection", author = "Edward Rosten and Tom Drummond", year = "2006", month = "May", booktitle = "European Conference on Computer Vision (to appear)", notes = "Poster presentation", url = "http://mi.eng.cam.ac.uk/~er258/work/rosten_2006_machine.pdf" }