Addressing the challenge of efficiently managing, searching, and analyzing large volumes of video content in the digital age, where traditional methods are time-consuming and ineffective. QMapper aims to solve this by leveraging AI to streamline and enhance the process of video analysis and digital asset management.
Object tracking is distinct from label detection in that it focuses on identifying and locating individual objects within a video frame, using bounding boxes and labels. For instance, in a video of vehicles at an intersection, object tracking would highlight and label each item like "car," "truck," or "bike," with bounding boxes and time-stamped segments showing their positions and movement over time.
The AI Face Analyzer feature identifies faces in videos, providing segments where faces are present. It optionally offers bounding boxes for each detected face and can also detect attributes like headwear, eye visibility, glasses, and expressions when enabled in the FaceDetectionConfig.
Speech Transcription converts spoken audio from videos into text, returning transcribed text blocks. It primarily supports English (US). Features include alternative words, profanity filtering, transcription hints, audio track selection, automatic punctuation, and speaker diarization to distinguish between multiple speakers.
The Logo Recognition feature detects, tracks, and recognizes over 100,000 brands and logos in videos, quantifying brand presence by counting appearances and measuring on-screen duration.
Text Detection uses Optical Character Recognition (OCR) to identify and locate text within video frames or segments, providing the text content along with its frame-level position and timestamp. Useful for media and entertainment, it can extract text like cast lists or subtitles and supports 60+ languages.
Explicit Content Detection identifies material in videos that may be unsuitable for minors, typically those under 18. This includes a range of content such as nudity, sexual acts, and explicit imagery. The technology also detects similar themes in animated or anime-style content.
Label detection identifies and annotates various entities like objects, locations, and activities in video frames, providing labels such as "train" or "transportation". Each label comes with a time segment and an entity ID for further exploration, unlike object tracking which focuses on individual objects within bounding boxes.
Shot change detection marks segments in a video where abrupt visual transitions occur. Each segment starts with a frame signifying a sudden change in the shot, distinguishing it visually from the preceding frame.
Person Detection feature identifies humans in videos, tracking them with bounding boxes. It not only detects people but also identifies specific body parts and characteristics like clothing color and type, providing detailed annotations for each detected individual.
Sign up to hear from us about updates