AI process thousands of videos?! - SAM2 deep dive 101

[@xXWillyxWonkaXx]

Jason you are a marvel mate. Punctual. Very knowledgeable. Delivers information that is truly valuable for the average "joe" for the community. Truly a marvel mate.

[@J3R3MI6]

Dude this was perfect. Please more SAM2 experiments. This tech is crazy powerful

[@Joe-bp5mo]

Pretty crazy how far we went - have anyone tried to build an AI based img/video search yet?

[@如日张]

you may index the video and timestamp with embedding text and use vector to search the videos

[@aidasliaudanskas7277]

Twelve labs

[@jasonfinance]

Thanks for explaining new models in depth with showcase of the building process

[@gabrieleguo]

Always bringing top notch content Jason, I really appreciate it, keep it up!

[@stevecoxiscool]

I am wanting to have an app that runs locally, which builds a library of all my saved youtube videos (cataloging), then lets me use a text prompt to not only find the videos that match the prompt, but also plays the videos at the exact time frame the data was matched. I save TONs of videos (Ctrl-D) in youtube like this one, but can't ever find them again !!!! LOL

[@DiegoSilva-dv9uf]

Valeu!

[@stevecoxiscool]

Very nice. I didn't see the use of the sam models since the missing ingredient was having another model to do object detection. Thanks !!!

[@RishabhMathur06]

@AIJasonZ please do more videos related to Florence-2 + SAM2 and create amazing real world solutions. Combining both may be revolutionary! Thanks!

[@MrRaveHaven]

I agree with everyone else - well done! This video is just what we needed right when we needed it.

[@tylertubridy3621]

Jason, you are a gentleman and a scholar

[@VivekPayasi]

Jason, thanks for this video. I actually need your guidance as i'm trying to deploy sam2 model for hundreds of cctv streams (reading frame by frame). Currently i'm using DeepStream to process these but the accuracy of detection and tracking is not that great with yolo models. Hence I want to try sam2 with florence VLM. If you have experience with this, can you please tell me if it is possible to integrate sam2 + florence VLM with Deepstream?

[@RobotechII]

Genius example, simple but powerful.

[@airevolution23]

Great information and very well explained!

[@PreciousToyo-dw2nj]

Please drop link to the Google colab notebook

[@ZombieRPGee]

Interactive element detection is going to be what I'm building!

[@courtenay100]

Would love a tutorial showing how to build a model that monitors a store and tracks customers and objects within the store to monitor foot traffic and other behaviors. 🙏👍

[@lordareello8221]

What are the system requirements

[@VisionCS2]

I smelled another AI Jason video coming keeeeeen

[@orlandocastellanos9263]

how much does the api cost?

[@ex3aliber]

Awesome video!! Thanks

[@mosca204]

Are you not going to share your code anymore?

[@sirishkumar-m5z]

With SmythOS, handle hundreds of videos with ease! This application makes it easy to manage large quantities and scale your content strategy. #SmythOS #AIVideo #ContentCreation

[@Itaivarochik]

Perfect 👌🏼

[@fx-studio]

All predeveloped and on a roll out. Mil is 10 years ahead at this point.

[@Username56291]

can you be my advisor please? im starting

[@kievisual]

Bro only makes videos when the sponsor rolls in :(

[@ZombieRPGee]

how dare he want to make money from his job :(

[@kievisual]

@@ZombieRPGee calm down