1. Pedestrian Monitoring

  

Frame 111                          Frame 152 

 

        Frame 165                      Frame 195

Figure 1. Mpeg 7 Test sequence 2. Red bar shows the spatial motion activity in x direction; Yellow points show the attention samples. Point size indicates the confidence of this sample. This figure illustrates the ability of maintaining multi-modal motion activity spatially. Both motion attention regions emerge and split during and after the crossing of the subjects

You can download the test clip :  AVI Format (not that all AVI files need divx codec )

The original video clip : AVI Format

2. Traffic Monitoring

         

       

    Frame 3  N_A = 0          Frame 191 N_A =272       Frame 193 N_A =147

   

   Frame 203 N_A =345           Frame 212 N_A = 238        Frame 243 N_A =0

Figure 2 Traffic monitoring sequence. This figure illustrates the both spatial and temporal traffic activity inferred from motion experience. Blue points are sensor samples while yellow points are attention samples. Red bar shows the spatial traffic activity in x direction. It evolves according to the spatial experience. N_S number of sensor samples is set to 200. N_A number of attention samples changes each time according to the temporal experience.

You can download the test clip: AVI Format (needs divx codec )

The original video clip: AVI Format

Figure 3. motion attention saturation in the traffic monitoring clip.

Our method here can be used for monitoring the traffic status also. It shows that the attention is only aroused when the cars come. At other times, when Attention saturation ASat is zero, there are no attention samples. During this time, the only processing and analysis done is the sensor sampling.

(a) Frame 76      (b) Frame 81    (c) Frame 106   (d) Frame 120

(e) Frame 147     (f) Frame 152    (g) Frame 158  (h)Frame 268

Figure 4. Face detection sequence 1. (a) static frame N_A=0 (b)A chair moves N_A=414 (c)the chair stopped. N_A=0 (d) a person comes. N_A=791 (e) a person. N_A=791 (f) one person. N_A=791 (g) one person. N_A=791 (h) static frame. N_A=2.

 As shown in Figure 4, N_S number of sensor samples is set to 200.The number and spatial distribution of attention samples can dynamically change according to the face attention. In Figure 11(a), there is no motion in the frame, so N_A, the number of attention samples is zero. No face detection is performed. In Figure 4(b), when a chair enters, it alerts the motion sensor and attention is aroused. N_A increases to 414. Face detection is performed on the 414 attention samples. But the face detector verifies that there is no face there. In Fig 4(c) as the chair stops, there is no motion and so the attention samples vanish. In Figure 4(d)-(h) attention samples come on with the face until the face vanishes.

You can download the captured test sequence: MPEG Format or AVI Format (needs divx codec )

2. Multiple Camera Surveillance

1. Multiple Camera Selection

(a)

(b)

Figure 5. Group 1 (a) camera position (b) The attention saturation of surveillance videos

        Frame 10 from Camera 1(default Camera)    Frame 160 shift to Camera 2 ASat: 241.8

    Frame 283 shift to Camera 3  ASat: 279.8      Frame 640 shift to Camera 2 ASat: 127.7

Figure 6. The sequence of the Main monitor in Group 1

You can download the Main monitor of Group 1  sequence:  AVI Format (needs divx codec )

(a)

(b)

Figure 7. Group2 (a) camera position (b) The attention saturation of surveillance videos

 Frame 13: Camera 1(default camera) ASat: 810.3   Frame 65: Shift to Camera 3 ASat: 357.3

Frame 272: Shift to Camera 2 ASat:120.7          Frame 682: Camera 1 ASat: 357.3

Figure 8. The sequence of main monitor in Group 2

You can download the Main monitor of Group 2  sequence:  AVI Format (needs divx codec )