170 likes | 318 Views
Faster and More Accurate Face Detection on Mobile Robots Using Geometric Constraints. Michael Dixon, Frederick Heckel , Robert Pless , William D. Smart Washington University in St. Louis. Motivation. Why is object detection important? Examples Autonomous vehicles Human-robot interaction.
E N D
Faster and More Accurate Face Detection on Mobile Robots Using Geometric Constraints Michael Dixon, Frederick Heckel, Robert Pless, William D. Smart Washington University in St. Louis
Motivation • Why is object detection important? • Examples • Autonomous vehicles • Human-robot interaction
Object detection in 2D • Common approach • Learn a classifier from labeled examples • Exhaustively classify all subwindows in the image Over 3 million subwindows in a 640×480 image
Object detection on a robot • Advantages of robot vision • Known camera position and calibration • Knowledge of environment • Additional sensors • Goal: use this additional information to reduce unnecessary tests by the detector • Less computation • Fewer opportunities for false positives
Images are projections • Image subwindows correspond to a range of possible 3D positions and sizes If certain 3D positions and sizes can be ruled out, you can avoid testing the corresponding subwindows
Geometric constraints • Establish bounds on an object’s physical position and size • Use the known camera to relate those physical bounds to the image • Only test a subwindow with the detector if it satisfies the geometric constraints
Face detection on robots • Perfect for geometric constraints • Robot has known camera • People’s heights and sizes are constrained Known calibration Known position Height constraint Size constraint
Applying geometric constraints • For each subwindow, compute the range of depths consistent with the constraints • If the range is empty, the subwindow can be safely ignored Satisfies height constraint Satisfies size constraint
Height constraints • Given: • height of camera, h • subwindow center, (u,v) • ground-plane normal, n • height constraints, hmin and hmax • Project (u,v) to ray in world space, r • Compute the range of depths consistent with the height constraints r hmax h hmin n
Size constraints • Given: • width of the subwindow, w • the camera’s focal length, fx • size constraints, smin and smax • Compute the range of depths consistent with the size constraints smax rs smin
Combining constraints • For each subwindow, (u,v,w): • Compute range of valid depths, • Can pre-compute quickly • Assuming no camera roll,u can be ignored • For all v, compute Dh • For all w, compute Ds • Store each D in v by wlook-up table • Only update if the camera tilts or zooms Ds(w) Dh(v) D
Incorporating depth measurements • For each subwindow, compute range of depths, M, consistent with external depth measurements • If D and M do not overlap, the subwindow can be safely ignored Stereo disparity
Incorporating laser range data • Project laser readings into image • Estimate depth at each pixel assuming a uniform footprint • Preprocess the range data to fill small gaps Laser rangefinder
Evaluation • B21r, SICK PLS Laser rangefinder (180 degrees), Bumblebee stereo camera • Control: Ran OpenCV detector at 25 scales from 20 to 200 pixels, scaling by a factor of 1.1, shifting window in increments of 0.5·w/w0 300 pairs of stereo images, 416 labeled faces
Reduction in computation • Geometric constraints alone cut computation in half • Incorporating depth from stereo or laser reduces computation by an average of 85%
Improved accuracy • Evaluated the detector over the full range of sensitivity thresholds • Three times fewer false positives
Example results Control • OpenCV face detector • Geometric constraints + laser measurement Laser