How Image Watermarking Works – Computerphile
Episode details
Episode description
Image watermarks are a way to embed information into a picture, Dr Mike Pound takes us through Adobe's Trust Mark to explain the idea & covers some of the pros and cons. 00:00 - Introduction: Watermarking Images vs Text 00:28 - Steganography vs Watermarking & Metadata Limitations 01:39 - Why Least Significant Bits (LSB) Fail 04:04 - Deep Learning for Watermarks (Encoder-Decoder) 05:37 - How Do We Detect & Read Embedded Codes? 07:18 - Robust Watermarking: Training with Augmentations 09:21 - Handling Any Image Size (Trustmark & 256px Base) 11:36 - Live Demo: Watermarking an Image in Python 13:40 - Stress Testing: Compression & Resizing 15:48 - Extreme Edits: Cropping & Color Shifts 17:23 - Adversarial Attacks: Copying Watermarks Between Images 19:39 - AI Tracking, Provenance, & Digital Signatures Computerphile is supported by Jane Street. Learn more about them (and exciting career opportunities) at: https://jane-st.co/computerphile This video was filmed and edited by Sean Riley. Computerphile is a sister project to Brady Haran's Numberphile. More at https://www.bradyharanblog.com
Watch also from the same channel
More episodes from this channel
You might be interested in
