A historical look at mid-2000s object tracking: state estimation, appearance models, detection, data association, geometry, and identity-aware evaluation.
How Early Statistical Speech Recognition Turned Sound Into Words
A historical guide to early statistical speech recognition, covering HMMs, GMMs, lexicons, language models, adaptation, evaluation, and data scarcity.
Reconstructing Speech Technology for Low-Resource Languages
How low-resource speech systems are built from recordings, transcripts, linguistic knowledge, evaluation, and community-led data governance.
Language Identification in the 2000s: Character N-Grams, Speech Models, and Difficult Cases
A historical look at 2000s language identification, covering character n-grams, speech-based LID, encoding issues, code-switching, under-resourced languages, and evaluation.
How Conference Proceedings Became Digital Archives in the Mid-2000s
How mid-2000s conference proceedings evolved from printed event records into searchable digital archives through metadata, persistent identifiers, indexing, and preservation.
Early Document Processing: OCR, Layout Analysis, and the Limits of Recognition
A historical look at early document processing, from OMR and MICR to OCR, layout analysis, handwriting recognition, quality measurement, and archival retrieval.
How Digital Presentation Technology Changed Pattern Recognition Conferences in 2005
By 2005, laptops, digital projectors, multimedia demos, and CD-ROM proceedings were changing how speech and computer vision research was presented, evaluated, and preserved.
Biometric Recognition in the Mid-2000s: Systems, Errors, and Deployment Lessons
How mid-2000s biometric systems handled fingerprints, faces, irises, voice, standards, error rates, and the practical limits revealed by deployment.
A Brief History of Speech Synthesis: Machines, Models, and Language Communities
Trace the history of speech synthesis, from mechanical speaking machines and formant synthesis to concatenative and statistical voices, with attention to accessibility and under-resourced languages.
Why Dense Stereo Reconstruction Returned to the Forefront in the Mid-2000s
How dense and semi-dense stereo methods, optimization, calibration, and benchmarks renewed stereoreconstruction research in the mid-2000s.