CS 584 Assignment 1, part 5

For part 5, two images were processed (mine and Dr. Boult's). Each image was blurred, random noise was added, then the face was detected and the location of the detection rectangle was recorded. This was done several times with different noise/blur factors applied. The first rectangle recorded represents the baseline rectangle (pure image, no distortions).

From the data and the pictures, you can see that with Dr. Boult's picture, there are times where the detection algorithm picks the oval object in the background over Dr. Boult's face. In the results from my picture, there are no distracting objects so the recognition software gets it "right" every time. The location error is simply the distance from the center of the rectangle in question to the baseline rectangle. In the data for Dr. Boult, results from failed detections are highlighted.

Results using Dr. Boult's image
Successful detection Unsuccessful detection
Data x y width height location error
Results: blur=0, noise=0 154 76 216 216 0.00
Results: blur=3, noise=2 32 114 70 70 127.78
Results: blur=3, noise=4 146 74 228 228 8.25
Results: blur=3, noise=8 148 74 224 224 6.32
Results: blur=3, noise=12 154 78 214 214 2.00
Results: blur=3, noise=16 156 80 212 212 4.47
Results: blur=5, noise=2 148 68 232 232 10.00
Results: blur=5, noise=4 146 72 228 228 8.94
Results: blur=5, noise=8 156 76 214 214 2.00
Results: blur=5, noise=12 154 76 218 218 0.00
Results: blur=5, noise=16 150 76 222 222 4.00
Results: blur=7, noise=2 158 76 208 208 4.00
Results: blur=7, noise=4 154 76 220 220 0.00
Results: blur=7, noise=8 152 76 220 220 2.00
Results: blur=7, noise=12 28 106 76 76 129.52
Results: blur=7, noise=16 146 74 226 226 8.25
Results: blur=9, noise=2 154 74 222 222 2.00
Results: blur=9, noise=4 154 80 216 216 4.00
Results: blur=9, noise=8 150 76 222 222 4.00
Results: blur=9, noise=12 148 74 224 224 6.32
Results: blur=9, noise=16 148 76 224 224 6.00
Results: blur=11, noise=2 164 76 210 210 10.00
Results: blur=11, noise=4 156 74 212 212 2.83
Results: blur=11, noise=8 32 114 70 70 127.78
Results: blur=11, noise=12 146 72 230 230 8.94
Results: blur=11, noise=16 142 74 232 232 12.17
Results: blur=13, noise=2 150 72 216 216 5.66
Results: blur=13, noise=4 146 70 232 232 10.00
Results: blur=13, noise=8 148 74 226 226 6.32
Results: blur=13, noise=12 146 74 230 230 8.25
Results: blur=13, noise=16 148 72 228 228 7.21
Results: blur=15, noise=2 40 232 120 120 193.21
Results: blur=15, noise=3 44 104 48 48 113.51
Results: blur=15, noise=4 152 72 226 226 4.47
Results: blur=15, noise=8 148 72 228 228 7.21
Results: blur=15, noise=12 146 74 228 228 8.25
Results: blur=15, noise=16 144 72 232 232 10.77

 

Results using my image
Typical detection
Data x y width height location error
Results: blur=0, noise=0 112 60 136 136
Results: blur=3, noise=0 112 60 138 138 0.00
Results: blur=3, noise=1 112 60 138 138 0.00
Results: blur=3, noise=2 114 62 136 136 2.83
Results: blur=3, noise=5 114 62 134 134 2.83
Results: blur=3, noise=6 108 60 142 142 4.00
Results: blur=3, noise=7 110 62 138 138 2.83
Results: blur=3, noise=8 110 60 138 138 2.00
Results: blur=3, noise=9 112 62 134 134 2.00
Results: blur=5, noise=2 104 52 150 150 11.31
Results: blur=5, noise=5 112 62 136 136 2.00
Results: blur=5, noise=9 112 60 136 136 0.00
Results: blur=7, noise=2 106 56 144 144 7.21
Results: blur=7, noise=5 112 60 138 138 0.00
Results: blur=7, noise=9 112 60 138 138 0.00
Results: blur=9, noise=2 110 60 146 146 2.00
Results: blur=9, noise=5 108 60 144 144 4.00
Results: blur=9, noise=9 110 62 140 140 2.83
Results: blur=11, noise=2 104 56 144 144 8.94
Results: blur=11, noise=5 112 62 134 134 2.00
Results: blur=11, noise=9 114 64 128 128 4.47
Results: blur=13, noise=2 112 60 138 138 0.00
Results: blur=13, noise=5 112 62 136 136 2.00
Results: blur=13, noise=9 114 64 130 130 4.47
Results: blur=15, noise=2 112 62 130 130 2.00
Results: blur=15, noise=3 110 58 136 136 2.83
Results: blur=15, noise=4 110 62 138 138 2.83
Results: blur=15, noise=5 116 64 130 130 5.66
Results: blur=15, noise=9 114 64 128 128 4.47