Built from the highest quality images available, DMDF Faces V1 combats the next generation of deepfake threats and harmful misinformation for free.

SAN FRANCISCO, CALIFORNIA, UNITED STATES, November 3, 2022 /EINPresswire.com/ -- DeepMedia, a pioneer in the creation and detection of synthetic media for dual-use commercial and defense purposes, today announces the release of DeepMedia DeepFake (DMDF) Faces V1, the first publicly-available dataset built to detect the next generation of advanced deepfakes—trained from the highest quality and widest range images available.

DMDF Faces V1 is the most powerful ready-to-use toolkit for the creation of advanced and accurate deepfake detectors today, available to researchers for free for the first time. Most importantly, this dataset solves the major issues that plagued previous datasets: low-quality training data and lack of diverse training sets.

Consisting of three different alignment methods of authentic images and eight total fake datasets, DMDF Faces V1 provides the most robust training data ever available in one place. DMDF Faces V1 includes hyper-realistic synthetic media of over 100K identities combining data from more than seven different deepfake creation techniques. Utilizing five unique cropping techniques along with original audio analysis, the dataset includes two sets of high-accuracy facial landmarks for each video frame to aid in multi-modal detection.

“We’re incredibly excited to release this free dynamic toolkit to help everyone safeguard against ongoing deepfake threats and help make the world a safer place,” said Rijul Gupta. “As dangerous and realistic deepfakes continue to spread, our society needs accurate and accessible detectors like DMDF Faces V1 to protect truth and ethics.”

To support the release of the dataset, DeepMedia is also announcing the launch of its Twitter bot, which can analyze videos on social media in real time and provide a perspective on if the media is real or a deepfake. All users need to do is respond to a video and tag @DeepMedia_AI, and their bot will provide a response based on the dataset.

The release of this industry-defining dataset and detector bot is just the latest resource DeepMedia has provided for the public. Earlier this year, DeepMedia released their DM-AER-DeepFake-V1 dataset to detect deepfaked aerial satellite imagery threats to global intelligence at the highest level, trained on over 1M images of synthetic aerial images and 120K real aerial images. DeepMedia has also recently briefed several defense and intelligence communities on deepfake detection and was recently selected for a phase 2 SBIR grant to assist in creating advanced deepfake detection tools for the AFRL. Last month, DeepMedia led a free public workshop on deepfake detection in partnership with MGMWERX, an organization created under a partnership between DEFENSEWERX and the Air Force Research Laboratory, that educated attendees on deepfake detection tools and the roles that public and private datasets play in the next wave of Generative AI. Additional information can be found here.

DeepMedia is continuously pursuing more opportunities with additional end-users to advance research, find solutions and evolve their detection technology. V2 of this dataset is coming soon and will include additional data on synthetic facial manipulation networks such as Thin-Plate-Spline, StarGAN3, and others.

Those interested in viewing and testing DMDF Faces V1 can access it here: https://github.com/deepmediaAI/DMDF. For more information, please visit deepmedia.ai

About DeepMedia:
DeepMedia is a San Francisco-based synthetic media platform company that pioneers industry-leading generation and detection capabilities for dual-use commercial and defense purposes. The company is an expert in synthetically-manipulated audio/video (e.g., deepfakes) detection.

DeepMedia offers ethical-only generation services such as its proprietary platform, DubSync, which is currently in production with the United Nations, film & TV production studios, and the biggest content-creators on YouTube to automatically translate and dub videos in 10+ languages. As a leader in the rapidly developing synthetic media space, DeepMedia aims to provide the Defense and Intelligence Communities with unparalleled access to technology and data to ensure that the United States and its allies remain dominant in all forms of synthetic generation and detection. For more information, visit https://www.deepmedia.ai/.

Greg Russo
DeepMedia
+1 732-241-7860
deepmedia@derris.com

Legal Disclaimer:

EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.