Testing outdoor markerless tracking with FAST/SURF feature detector.
The plane of the camera is not parallel to the earth, that make difficult for eye to estimate precision.
Feature detected with multistage FAST and fitted with SURF descriptors
Less strict threshold give a lot more correspondences, but also some false positives
Just checked GLBenchmarks for Samsung INNOV8.
Samsung GT-i8510 Innov8 performance details:
High-level 3D Performance
GLBenchmark HD ES 1.0 : 1602 Frames
GLBenchmark HD ES 1.1 : 628 Frames
GLBenchmark Pro ES 1.0 : 328 Frames
GLBenchmark Pro ES 1.1 : 35 Frames
CPU Performance: Float : 988
CPU Performance: Integer : 4183
For comparison –
Nokia N95 8GB performance details:
High-level 3D Performance
GLBenchmark HD ES 1.0 : 1944 Frames
GLBenchmark HD ES 1.1 : 1766 Frames
GLBenchmark Pro ES 1.0 : 395 Frames
GLBenchmark Pro ES 1.1 : 570 Frames
CPU Performance: Float : 1411
CPU Performance: Integer : 4254
Old Nokia N95(8Gb) still beat best of newish Symbian devices hands down.
Other platforms had 600+ Mhz devices long time ago, Palm switched to new generation CPU – OMAP3430. After two years no one still able to produce better high-performance Symbian device ?
Experimenting with multiscale FAST detector with images from cell phone camera.
so far so good…
Polynesian Stick Charts were completely different way of navigation, they were mapping not only locations, but also oceanic swells, patterns of waves.
Specific map encoding was closely guarded secret, known only to group of navigators who own them.
Navigating by the wave pattern navigator “would crouch in the bow of his canoe and literally feel every motion of the vessel.”They “concentrated on refraction of swells as they came in contact with undersea slopes of islands and the bending of swells around islands as they interacted with swells coming from opposite directions.”
Fascinating staff, kind of technology which could have been developed by alien, or in alternate history line.
Retweet from @bruces
It seems a new evidence found, that our universe is not 3d but 2d surface. According to Holographic principle in in string theory physic inside some volume of space could be completely described by the theory restricted to the boundary surface of that volume. That is tree-dimensional physical processes inside the volume of space could be seen as illusory projection of the two-dimensional processes on the surface of that volume. The new development here is that as in quantum theory space-time should be granulated, that granulation should be primarily two-dimensional, on the surface of the volume. The space-time granulation inside the volume would be projection of the surface granulation. And – here the news – that projected granulation should be bigger than original surface granulation(plank length). That bigger granulation should create additional blur, or noise in physical effects. And it seems that noise exactly like that was detected at GEO600 gravitation waves detector
Sony DSG-G3 camera run busybox linux and have scene(?) and face recognition. With more and more computer vision algorithms integrated into cameras, and more powerful CPUs, cameras becoming essentially AR platform.
Testing FAST feature detector on the Mikolajczyk ’s dataset. Here scale space seems actually useful. With “brick wall” dataset repeatability goes form .3 to .7 with scale from 0 t 2^^3, and threshold/barrier lowering from 40 to 20.