A sketch like

1. void setup()
2. {
3.   size(1024, 720);
4. }
5.  
6. void draw()
7. {
8.   background(0);
9.  
10.   ellipse(frameCount % width, height / 2, 50, 50);
11. }
12.  
13. void keyPressed()
14. {
15.   exit();
16. }

works for me, ie. no need to put the focus on the frame to have keyPressed() to work, on Windows 7.
So perhaps the problem is elsewhere. You should do like me, starting from a simple sketch, building it up to see when it stops working as you want. It might be the renderer, the behavior of the video library or whatever.

BTW, moved from Programming Questions because there is, precisely, no code to look at.

Sorry, here is the code.
I created this program with the help of some talented users of this forum.
It permits to select photographies with the body, by moving left or right in front of the webcam.
When I start the program, I first have to click on the frame of the program and than click on Space.
If I don't do this, the program doesn't do anything when I press Space.

1. import processing.video.*;
   
   int numPixels;
   int[] backgroundPixels;
   int[] hist;
   int[] histTop;
   Capture video;
   int pressed=0;
   float presenceSum=0;
   //Dati relativi al gioco
   PImage bigImage;
   PImage big1;
   PImage big2;
   PImage big3;
   PImage big4;
   PImage big5;
   PImage big6;
   PImage small1;
   PImage small2;
   PImage small3;
   PImage small4;
   PImage small5;
   PImage small6; 
   PImage cursor;
   
   
   
   void setup()
   {
     size(800, 600, P2D);
     // Change size to 320 x 240 if too slow at 640 x 480
    
   
     video = new Capture(this, width, height);
     video.start();
     numPixels = video.width * video.height;
     // Create array to store the background image
     backgroundPixels = new int[numPixels];
     // Make the pixels[] array available for direct manipulation
     loadPixels();
   
     big1 = loadImage("0big.jpg");
     big2 = loadImage("1big.jpg");
     big3 = loadImage("2big.jpg");
     big4 = loadImage("3big.jpg");
     big5 = loadImage("4big.jpg");
     big6 = loadImage("5big.jpg");
     small1 = loadImage("0small.jpg");
     small2 = loadImage("1small.jpg");
     small3 = loadImage("2small.jpg");
     small4 = loadImage("3small.jpg");
     small5 = loadImage("4small.jpg");
     small6 = loadImage("5small.jpg");
     cursor = loadImage("cursor.png");
   
   }
   
   void draw()
   {
     if (video.available()) {
       video.read(); // Read a new video frame
       video.loadPixels(); // Make the pixels of video available
       // Difference between the current frame and the stored background
   
       hist = new int[width];
       float maxval = 0;
       int maxposx=0;
       int maxposy=0;
   
   
       for (int i = 0; i < numPixels; i++) { // For each pixel in the video frame...
   
         // Fetch the current color in that location, and also the color
         // of the background in that spot
         color currColor = video.pixels[i];
         color bkgdColor = backgroundPixels[i];
   
         // Extract the red, green, and blue components of the current pixel’s color
   
         float currR = red(currColor);
         float currG = green(currColor);
         float currB = blue(currColor);
   
         // Extract the red, green, and blue components of the background pixel’s color
         // esegue la stessa operaizone di prima ma in maniera più veloce
         float bkgdR = red(bkgdColor);
         float bkgdG = green(bkgdColor);
         float bkgdB = blue(bkgdColor);
         // Compute the difference of the red, green, and blue values
         float diffR = abs(currR - bkgdR);
         float diffG = abs(currG - bkgdG);
         float diffB = abs(currB - bkgdB);
         // Add these differences to the running tally
         // Render the difference image to the screen
        
        if (diffR>20 &&  diffG>20 && diffB>20 && pressed==1) {
           pixels[i] = color(255, 255, 255);
           presenceSum = presenceSum + 1;
        }
        else {
   
    {
    pixels[i] = color(0, 0, 0);
    }
       }
     
       }
       updatePixels(); // Notify that the pixels[] array has changed
   
         textSize(30);
          fill (255);
         text("Assicuratevi che non ci sia nessuno davanti la webcam.", 5, 100); 
         text("Premere il tasto SPAZIO per far avviare il programma.", 5, 140);
        
   
       filter(ERODE);
   
   
       if (pressed==1 && presenceSum>width*height/100) {
         // Calculate the histogram
         for (int i=0; i<width; i++) {
           for (int j=0; j<height; j++) {
             if (  int(red(get(i, j)))  == 255 && int(green(get(i, j)))==255 && int(blue(get(i, j)))==255 )
               hist[i]=hist[i]+1;
           }
         }
   
   
         // Find the largest value in the histogram
         for (int i=0; i<width; i++) {
           if (hist[i] > maxval) {
             maxval = hist[i];
             maxposx = i;
           }
         }
         maxposy=height-int(maxval);
        
   
   
         if (key==' ') {
    
           background(0);
           image (cursor, width-maxposx-50, 40);
           image (small1, 100, 20);
           image (small2, 200, 20);
           image (small3, 300, 20);
           image (small4, 400, 20);
           image (small5, 500, 20);
           image (small6, 600, 20);
   
           if (bigImage != null)
             image (bigImage, 60, 120);
   
           if (width-maxposx>100&&width-maxposx<180)
           {
             bigImage =  big1;
           }
           else
   
               if (width-maxposx>200&&width-maxposx<280)
             {
               bigImage = big2;
             }
             else
   
                 if (width-maxposx>300&&width-maxposx<380)
               {
                 bigImage =  big3;
               }
               else
   
                   if (width-maxposx>400&&width-maxposx<480)
                 {
                   bigImage =  big4;
                 }
                 else
   
                     if (width-maxposx>500&&width-maxposx<580)
                   {
                     bigImage =  big5;
                   }
                   else
   
                       if (width-maxposx>600&&width-maxposx<680)
                     {
                       bigImage =  big6;
                     }
   
                     else
   
                     {
                       bigImage = null;
                     }
         }
         if (key=='v') {
           for (int i = 1; i < width; i++) {
             for (int j = 0; j < hist[i]; j++) {
               color c = color(255, 0, 0);
               set(i, j, c);
             }
           }
   
           for (int i = maxposx-3; i < maxposx+4; i++) {
             for (int j = 0; j < height; j++) {
               color c = color(0, 255, 0);
               set(i, j, c);
             }
           }
   
           for (int j = height-int(maxval)-3; j < height-int(maxval)+4; j++) {
             for (int i = 0; i < width; i++) {
               color c = color(0, 255, 0);
               set(i, j, c);
             }
           }
         }
       }
     }
   }
   
   
   void keyPressed() {
     if (key==' ') {
       video.loadPixels();
       arraycopy(video.pixels, backgroundPixels);
       pressed=1;
        }
     
   }
   
   
   

OK, moved here since it uses the standard library.
I don't have a camera, so I cannot test the code. Perhaps somebody else can confirm.
Indicating which platform you use can help too, perhaps this behavior is related to your system.
A solution might be to show a button "click here to start", perhaps...

Also, perhaps try:
frame.requestFocus();
in setup().

I use Windows 7 64bit.
I tried to insert frame.requestFocus(); but nothing changed. I still have to click on the frame.