Continue Reading

At-Scale OCR Of Television News Experiments: Results From A Sample Circa-2010 SD-Resolution Arabic-Language Broadcast

Earlier this week we demonstrated the results from applying GCP Cloud Vision to a grid montage of 1fps still frames…

Continue Reading

At-Scale OCR Of Television News Experiments: Two Interesting Examples Of The OCR & Inconsistencies Of OCR At Scale

Earlier this week we explored the ability of GCP's Cloud Vision API to robustly OCR a grid of 224 video…

Continue Reading

Snopes: Video Does Not Show Biden Wandering Off into Amazon Rainforest

Snopes uses the TV News Archive to verify that a clip seen on social media is authenticate, but used out…

Continue Reading

At-Scale OCR Of Television News Experiments: Results From A Sample Circa-2010 SD-Resolution English-Language Broadcast

Last week we explored how combining large numbers of video frames into a single grid-based montage allows us to vastly…

Continue Reading

Why Large Multimodal Models (LMM) Like ChatGPT Are Unsuitable For Production OCR

While Large Multimodal Models (LMMs) are increasingly positioned as replacements for the full range of classical AI systems like OCR,…

Continue Reading

Tomorrow: Don't Inform, Inflame: AI's Limitations

For those here at Web Summit 2024 in Lisbon, head to Stage 15 tomorrow to see Kalev's stage talk "Tomorrow:…

Continue Reading

This Week: Web Summit 2024 In Lisbon

Join Kalev at Web Summit 2024 in Lisbon this week, which is drawing more than 70,000 attendees from over 160…

Continue Reading

At-Scale OCR Of Television News Experiments: More Grid Layout Experiments

As we ramp up our work to expand OCR to the complete Internet Archive Television News Archive, the key to…

Continue Reading

Comparing The New Visual Explorer Intelligent Thumbnail Generator Vs The Old Fixed Thumbnails

The current Visual Explorer represents each television news broadcast using a single frame extracted exactly 60 seconds into the broadcast….

Continue Reading

The Fascinating Findings Of Applying Advanced OCR To Television News: Transcribing Projected Text

As we continue ramping up our experiments applying GCP's Cloud Vision API OCR to television news, it never ceases to…

Continue Reading

Next Week: Web Summit 2024 In Lisbon

Join Kalev at Web Summit 2024 in Lisbon next week, which is drawing more than 70,000 attendees from over 160…

Continue Reading

Can LMM's Like ChatGPT 4o And Gemini Yield Better Results Than Cloud Vision's Classical AI OCR?

Yesterday we explored the challenges of OCR'ing television news, in which a single change to one portion of an image…

Continue Reading

An OCR Tale Of Two Images: The Impact Of Small Changes On OCR Block Reconstruction

Classical production OCR systems work by recognizing text in an image and grouping sequences of related characters and words into…

Continue Reading

AFP Fact Check: Fabricated CNN Election Broadcast Spreads Online, Sparking False Cheating Claims

The AFP's Fact Check relies on the Internet Archive's TV News Archive to debunk a falsified purported screen grab from…

Continue Reading

Two Weeks: Web Summit 2024 In Lisbon

Join Kalev at Web Summit 2024 in Lisbon in two weeks, which is drawing more than 70,000 attendees from over…

Continue Reading

This Week Coverage Of The US Elections Has Turned Dark

A fascinating analysis by Richard Addy and Luba Kassova and their AKAS US Election Media Tracker examining the shift in…

Continue Reading

"Climate Change" Mentions On US 24/7 TV News Have Halved In The 2024 US Election Campaign Compared To 2020

A new analysis by the AKAS US Election Media Tracker examines coverage of climate change compared with the previous presidential…

Continue Reading

September 11th Fades On Television News & Web Searches

As the timeline below shows, mentions of 9/11 across television news from 2009 to present declined sharply from 2014 and…

Continue Reading

Mentions Of "Felon" On TV News Peaked In May And Never Rebounded

As the timeline below shows, mentions of "felon" peaked this past May, but the term has never rebounded, even in…

Continue Reading

OCR'ing Television News: Experiments With EasyOCR

Two months ago we evaluated the OCR performance of several models applied to television news frames, including dedicated SOTA OCR…

Continue Reading

Using Thumbnail Montages To Optimize AI-Based Video OCR Speed & Costs: Part 8 – Cost Comparisons

Looking back at the lessons we've learned over the past week optimizing the speed and cost of video OCR, why…

Continue Reading

Using Thumbnail Montages To Optimize AI-Based OCR Speed & Costs: Part 7 – What Have We Learned So Far?

What have we learned so far from our six-part series (Part 1, Part 2, Part 3, Part 4, Part 5,…

Continue Reading

Fortune: How The Democrat-Leaning News Media Is Unwittingly Aiding Trump

Luba Kassova and Richard Addy examine television media coverage of the 2024 presidential campaign. Read The Full Article.

Continue Reading

Using Thumbnail Montages To Optimize AI-Based OCR Speed & Costs: Part 6 – Further Grid Layout Experiments

Yesterday's experiments on how montage layouts impact OCR accuracy suggested that there were few differences between vertical, horizontal and grid-based…