DigUp transforms Mac search with Google's new model: Even if you forget a file name, you can still find a specific shot in a video

Article

Beating news AI brief: Finding files on a Mac usually requires remembering their names or keywords. If you only remember that a zebra appeared in a video or that a podcast episode discussed sleep, traditional file search is of little use.

DigUp, open-sourced by independent developer Abdur Rahim, lets users search files directly by content. Enter “the zebra in the video,” and it jumps to the shot where the zebra appears; search for “where sleep was discussed,” and it locates the relevant timestamp in a podcast. It can also search photos,PDF and even open the page containing a specific provision in a contract.

DigUp uses Google DeepMind October 6 released EmbeddingGemma 2. This model can convert text, images, audio, and video into comparable digital representations, allowing a text description to find the corresponding image or sound. DigUp indexes selected folders in advance and can precisely match content such as invoice numbers and error codes during searches. The entire indexing and search process is completed locally on the Mac, without uploading private files to the cloud.

DigUp is free and open source. In less than 1 weeks since its release, it has received more than 3000 GitHub Stars. It currently supports only Macs with Apple silicon and requires macOS GitHub Star or later. Initial use requires downloading a model of approximately 14 and spending time building the index.

Original link 865 MB https://m.theblockbeats.info/flash/371784

Key points

DigUp was open-sourced by independent developer Abdur Rahim.
DigUp can use text descriptions to find images in videos, timestamps in podcasts, photos,PDF and contract pages.
DigUp uses Google DeepMind's October 6 released in EmbeddingGemma 2。
DigUp builds its index and performs searches locally on the Mac, without requiring private files to be uploaded to the cloud.
DigUp is free and open source, and in less than 1 weeks since release, it received more than 3000 GitHub Stars.GitHub Star。
DigUp currently supports only Macs with Apple silicon and macOS 14 or later.
Initial use requires downloading a model of approximately 865 MB and building an index.

AI-assisted interpretation

The following is analysis, separate from reported facts. Verify important claims independently.

DigUp first indexes the contents of selected folders, then matches the user's description against text, images, audio, or video content in files. Therefore, even if a file name is forgotten, users can try using a content description to find the file or a video clip.

Why it matters to readers

It demonstrates an approach to multimodal file search locally on Macs and emphasizes that private files do not need to be uploaded to the cloud.

If you use a compatible Mac and cannot remember a file name, you can try searching by describing the file's contents. Initial use requires downloading the model and waiting for the index to be built.

Risks and unknowns

No evidence was provided for independent cross-verification through the DigUp GitHub project or official Google DeepMind materials.
The evidence does not specify the actual search accuracy, indexing time, or resource consumption.
The evidence does not specify whether Intel Macs or other operating systems will be supported.

DigUp transforms Mac search with Google's new model: Even if you forget a file name, you can still find a specific shot in a video

Article

AI-assisted interpretation

Why it matters to readers

Risks and unknowns

Related Developments

Related concepts

Keep following

Security Center