Remaining Challenges talk - Informedia Digital Video ….ppt

NoD and Multilingual Status ReportApril 1998 Carnegie Mellon University Howard D. Wactlar;MLI and NoD Tasks;Accomplishments to Apr 98;Accomplishments to Apr 98 (cont’d);Accomplishments to Apr 98 (cont’d);Automated Abstraction and Summarization;“Na?ve” Poster Frame Result List (Uses First Shot Image);Query-based Poster Frame Result List;Query-based Poster Frame Selection Process;Enhances browsing and discovery over directed search Different methods from several areas being evaluated Information retrieval - vector space methods - relevance feedback Speech recognition - hidden Markov models Statistics - k-nearest neighbors - exponential models;KNN-based Topic Detection;Speech Recognition for Index Generation; Segmentation - Creating the Video Paragraph;Segmentation - Commercial Detection;Ad Removal based on Black Frame and Scene Change Detection;Segmentation - Language Models;A plot of the ratio of the two language models as a function of the relative position in a segment. ;Image component crucial to news corpus Capture of text overlayed on the video image Detected, filtered, OCR’d, incorporated into content and indexed;Video OCR Block Diagram;Video Frames (1/2 s intervals);Text Detection False Alarms;Text Detection Misses;Challenges for VOCR Preprocessing;VOCR Preprocessing Problems; ;Character recognition - 83%Word recognition - 70% Language model post processing will improve word recognition rate, but new names and places will not be in language model Important adjunct to Name-It: name/face correlation through co-occurrence matrices;Annotation fields contain metadata automatically derived from the content (e.g. topics, chyron) Annotations are included in the index (searchable separately or combined with transcript) Personal annotations are typed or spoken comments that are established on a per user basis bookmarking or commentary fully indexed and searchable with other data;Long-time concern about video fidelity on internet Compromi