Although facial recognition algorithms look for similarities when returning candidates in a search, significant facial changes such as weight gain or loss do affect matching accuracy. Most systems recognize age progression and certain changes in skin tone but it is important to note that large weight gains and losses do impact performance by reducing the matching accuracy overall.
In cases such as these, a match will most certainly never return in a top tier ranking as the possible match will more than likely reside deeper in the candidate list. To better assist with matching accuracy it is recommended the end user apply filters when applicable (gender, race, age, location, height, weight, or any other descriptive feature) to your searches. This narrows the candidates to levels of specificity and will provide better chances for the candidate to return because the algorithms are searching against a smaller defined dataset of faces.