Example 3: An odd preview
How can this be? Answer: this image was not an ‘instant’ result. Exposure was reduced very low, after TFOD had recognized the Duck.
The implementations of TensorFlow Lite (and Vuforia) are good at tracking a currently-identified object (or image) through translation, rotation, partial blockage, and even extreme changes in exposure.