Our solution Smart Video Thumbnails uses artificial intelligence to choose thumbnails that will maximize the CTR (appealing
image) and the completion rate (relevant image) for your video assets. We use cognitive engines to analyze the characteristics
of each frame of a video : face recognition, sentiment analysis, object detection, OCR etc. A machine-learning algorithm
trained on domain-specific datasets selects multiple relevant candidates that are tested in-vivo until the system
finds the best thumbnail among the candidates. A/B testing with the default video thumbnails is used to evaluate the
performance of our solution.