With this Code , I found out the possible Head and Torso from Sports Images Separately and connected them all by matching their orientation , so if Person x 's Face is Detected it will not map into Person y's Torso , it will map into Torso of x only irrespective of Torso Not Getting Detected , so with 100 % Accuracy , i detected Face and Torso
Here I have faced 4 issues / cases : a. Face and Body Detected b. Face Detected but Body not Detected c. Face not Detected but Body Detected d. Face and Body not Detected
a. Face and Body Detected : For this case we have no issue we will move to next step b.Face Detected but Body not Detected: For this we consulted a paper where it was mentioned height of body upto thighs is 7 times the height of face , so we expanded the face downwards towards opposite boundary upto 7*height of face length c.Face not Detected but Body Detected: Same as above we will move the body upwards towards opposite boundary until my total height is total height / 7 d.Face and Body not Detected: Here we use Boundary Growing Method of detecting Undetected face and then expanding towards down boundary to get the undetected body
Steps to Run :
- Run first torso_face_Combo.m 2.Then Run ConnectingHeadAndBody.m for potraying the 4 cases above
- Run isi.py to extract specifically Point (d)