Text this: Vision-Based American Sign Language Classification Approach via Deep Learning