5 Use Cases for GPT-4 Vision API (and DALL-E 3)

In today's video I show you some example use cases for the new GPT-4 Vision API
GitHub: github.com/unconv/gpt4v-examples
GPT-4 Browsing Video: • GPT-4 Vision API + Pup...
Support: buymeacoffee.com/unconv
Consultations: www.buymeacoffee.com/unconv/e...
Memberships: www.buymeacoffee.com/unconv/m...
00:00 1. Image categorization
04:38 2. Image laundering
06:33 3. Website scraping
08:24 4. Website monitoring
10:44 5. Sketch to App
14:57 Source code walkthrough
19:20 Outro

[@oscaroneill8222]

Amazing work! I loved seeing all of these examples in action.

[@HealthyHive-oo6vc]

15:19 Its really cool👌👌

[@FillTechmx]

Bro your content REALLY ROCKS, never find something like this

[@zeta_meow_meow]

BEST BEAST

[@TheVisitorX]

I find the topic of vision and image creation/image recognition super interesting. I work for an online pharmacy shop. I could easily imagine this topic for our customers too. For example, to simply scan an image and find the right products (or alternatives). It would also be interesting to use it to monitor certain products or actions. In general, it is probably quite good for OCR. Unfortunately, the issue of data protection rules in pharmacies is very strict, otherwise this could also be used to process prescriptions for example. But there are probably hundreds of other things you could do with it. I would be happy to see more about the topic of speech recognition/image recognition! I always find your videos very good as inspiration for my own projects!

[@unconv]

Yeah, I'm really waiting for the day when these models can be run cheaply on your own server.

[@ghazouaniahmed766]

am still waiting for signature recognition, extraction and identification. If that works will be great

[@pradipdutta7851]

What's the prerequisite to build this project?

[@hakancarlsson2788]

Very good video. Is it possible to maybe use the code for indexing your own videos (from images every 2 seconds) and search for a part like "man on beach" and get link and timecode to that clip as the answer?

[@unconv]

Should be possible, but would take quite a few API requests. I might try something like that. Thanks!

[@peakMomentum1]

If I am understanding the question correctly, you want time stamps to a video. You can simply download its transcript along with timelines and chatgpt can reply with link and timestamp to your exact query in the video