def get_bounding_box(keyword, url, filename): # get the image img = Image(url) # resize the image so things aren't so slow, if necessary w, h = img.size() if w > 1200 or h > 1200: maxdim = max(w, h) ratio = math.ceil(maxdim/800.0) print " resizing..." img = img.resize(w=int(w/ratio), h=int(h/ratio)) else: ratio = 1 # get the canvas disp = Display((800, 800)) # text overlay textlayer = DrawingLayer(img.size()) textlayer.setFontSize(30) cx, cy = 10, 10 for xoff in range(-2, 3): for yoff in range(-2, 3): textlayer.text(keyword, (cx + xoff, cy + yoff), color=Color.BLACK) textlayer.text(keyword, (cx, cy), color=Color.WHITE) # two points to declare a bounding box point1 = None point2 = None while disp.isNotDone(): cursor = (disp.mouseX, disp.mouseY) if disp.leftButtonUp: if point1 and point2: point1 = None point2 = None if point1: point2 = disp.leftButtonUpPosition() else: point1 = disp.leftButtonUpPosition() bb = None if point1 and point2: bb = disp.pointsToBoundingBox(point1, point2) elif point1 and not point2: bb = disp.pointsToBoundingBox(point1, cursor) img.clearLayers() drawlayer = DrawingLayer(img.size()) if bb: drawlayer.rectangle((bb[0], bb[1]), (bb[2], bb[3]), color=Color.RED) # keyboard commands if pygame.key.get_pressed()[pygame.K_s]: # skip for now raise Skip() elif pygame.key.get_pressed()[pygame.K_b]: # mark it as an invalid picture raise BadImage() elif pygame.key.get_pressed()[pygame.K_RETURN]: if point1 and point2: bb = disp.pointsToBoundingBox(scale(ratio, point1), scale(ratio, point2)) return bb elif not point1 and not point2: bb = disp.pointsToBoundingBox((0, 0), (w, h)) return bb drawlayer.line((cursor[0], 0), (cursor[0], img.height), color=Color.BLUE) drawlayer.line((0, cursor[1]), (img.width, cursor[1]), color=Color.BLUE) #drawlayer.circle(cursor, 2, color=Color.BLUE, filled=True) img.addDrawingLayer(textlayer) img.addDrawingLayer(drawlayer) img.save(disp)
# main loop # while not disp.isDone(): img = cam.getImage() img = img.flipHorizontal() # just for my convenience - simulates mirror :) if inVideoMode: (orig_h, orig_w) = (img.height, img.width) # required for VideoStream px = disp.mouseX py = disp.mouseY left_down = disp.leftButtonDownPosition() # grab coordinates for event press left_up = disp.leftButtonUpPosition() # grab coordinates for event release if left_down is not None: # left mouse button pressed # start drawing rectangle (lx, ly) = left_down if left_up is not None: # left mouse button released # stop drawing rectangle, go into zoom mode (lx, ly) = (None, None) inZoomMode = True if not inZoomMode else False # enter zooming mode zoom = (sx, sy, w, h) if inZoomMode: img = img.crop(zoom[0], zoom[1], zoom[2], zoom[3])