3 ms·
Show HN: Semantic Search of 1000 Top Movies of All Time
- mutant 2y agohurt Locker is apocalyptic? No thanks.
- pjsg 2y agoI tried entering "Marty Feldman" as a search query. Unfortunately, the responses don't show movie titles, but I couldn't spot him in the few that I tried. "The Sting" was one movie that I could identify. Needs more work.
- Beefin 2y agothese are movie trailer concepts, not titles or scenes. we ran the movies through a video embedding model: https://learn.mixpeek.com/vuse-v1-release/ https://learn.mixpeek.com/vuse-v1-release/
- deanputney 2y agoIt's not clear what can be searched on this. The interface suggests "concepts", and I'm not sure how I would define those in a movie trailer. Being able to search actors or movies would be interesting, but IMdB already serves this purpose. Searching dialogue is what I was hoping for, similar to morbotron.com or frinkiac.com. filmot.com does this for general YouTube videos, but I would love to see that applied to a quality film collection.
- Beefin 2y agoyeah theres millions of movie search by metadata (title, actors, even script) we're not doing that here... we've run movies through embedding models to capture the semantics and make them searchable. reproducible code: https://github.com/mixpeek/use-cases/tree/master/movie-trailer-analysis https://github.com/mixpeek/use-cases/tree/master/movie-trail...
- NBJack 2y agoAt a glance, the semantics seem kind of weak. It's hard to tell in many cases why it picked certain scenes in lieu of the concept searched.
- Beefin 2y agocan you give an example of some queries that failed?