Articulo Medium: Parte5d

get_results.py

raw_results = model.predict(images)
results = {}
labels = {0: 'cat', 1: 'dog'}
text_to_show = {'cat': 'meowww', 'dog': 'bupp'}

def drawText(im, text, color):
	text_size = cv2.getTextSize(text, cv2.FONT_HERSHEY_SIMPLEX, 1, 2)[0]
	text_pos_x = (im.shape[1] - text_size[0]) // 2
	text_pos_y = (im.shape[0] + text_size[1]) // 2

	cv2.putText(im, text, (text_pos_x, text_pos_y), cv2.FONT_HERSHEY_SIMPLEX, 1, color, 2)

for i in range(raw_results.shape[0]):
	if(raw_results[i] > 0.5): # Revisar en el Labels de training
		drawText(real_images[i], text_to_show[labels[1]], (0, 255, 0))
		results[path_images[i]] = (labels[1], raw_results[i])
	else:
		drawText(real_images[i], text_to_show[labels[0]], (0, 0, 255))
		results[path_images[i]] = (labels[0], 1 - raw_results[i])

	cv2.imshow('Image', real_images[i])
	cv2.waitKey(0)
	cv2.destroyAllWindows()
	
for path_im, (name, confidence) in results.items():
	print('{0} = {1} ({2})'.format(path_im, name, confidence))