An amusement park wants to capture, monitor, and store real-time video from cameras at several entrances. Security staff need immediate viewing access; stored footage must be indexed and available to the data science team. Which is the most cost-effective solution that satisfies these requirements?

The correct answer is to use Amazon Kinesis Video Streams with its built-in HLS capability. Kinesis Video Streams is designed for ingesting, processing, and storing video streams, and it automatically indexes the footage. Its HLS feature allows security staff to view live and archived video with low latency, fulfilling the real-time viewing requirement. This is also a cost-effective solution as it leverages a single service for multiple requirements. Using Amazon Kinesis Video Streams with Amazon Rekognition (first incorrect option) would add unnecessary cost if the primary requirement is just viewing, not advanced video analysis. Amazon Rekognition Video with GStreamer and Kinesis Data Streams (third incorrect option) is not ideal because Kinesis Data Streams is for data records, not optimized for video, and GStreamer adds complexity. Amazon Kinesis Data Firehose (fourth incorrect option) is for delivering streaming data to destinations like S3 or Redshift, not for direct video ingestion and real-time viewing with HLS.