|
Sorry if this has been answered somewhere else, but I feel like I'm kinda of in the deep end and don't know where to start. Lim looking to work on a script to create product tags for some shopify listings, and an image recognition API seems like it would be ideal. I know gpt-4 has some image recognition in it, but access seems a bit weird/inconsistent and might be overkill. What I want to do, is input a product title with an image (it's 3d printed miniature,so "vampire lord", and a pic of the sculpt), and have it spit out a list of product tags for my search/profuct-categories. I spend a ton of time indexing search terms, so some API would save me a ton of time.
|
#
¿
Apr 16, 2023 18:59
|
|
|
|
| # ¿ May 17, 2024 01:58 |
|
|
cinci zoo sniper posted:To clarify, you have a collection of tags, and for each pair of title+photo you want an automated way to select the best matching tags from the collection? Sort of. so Shopify works off collections, which for me would be broad categories of sculpts (undead, human, beasts,terrain, etc), with some sub-categories (undead would have vampires, ghosts,zombies etc) Each product has tags which describe the product which I can use to pop an item into a category, but are also used for searchability. I want it to output any matching tags I've made so a listing can be assigned to any relevant categories, but also potentially add some tags I haven't thought of for searchability. The second part might be overkill, may just make more sense to generate a huge list of predefined tags to constrain it Ex: I have a vampire castle sculpt. I want it to run through a tool, and output that it has the tags undead,vampire, and terrain. It would then further give me a handfull of adjacent tags like Dracula, Transylvania, count etc for search Edit:I feel like I would need to train my own model? Given I want it to know what a mini of a vampire looks like. I have titles tied to 1.5k+ miniatures, but I feel like I might need to do something more broadly unless there's an existing solution w00tmonger fucked around with this message at 19:58 on Apr 16, 2023 |
|
#
¿
Apr 16, 2023 19:28
|
|
|
cinci zoo sniper posted:So, this is 2 separate tasks – 1) get tags for image, 2) generate new tags based on existing tags. The latter is something you can credibly do with any text model starting with GPT-3, after some prodding. The former, in industry terminology, would be “image classification”, assuming you have clean photos where the miniature is the only “feature” of the image. GPT-4 is multimodal, and accepts image inputs, but the image ingestion API is not enabled at the moment in the public OpenAI service, and it may ultimately depend on access to the 32k context length model, which is not publicly available eitehr, as yet. So you may have to shop around for some other model, and either consider trainign your own right away, or investigating fine-tuning an existing model with your data at a later point. Yeah I did a bit more digging around and 2 definitely seems easier/more-secondary. If I can manage to build out a dataset for a model, I might be able to use tensorflow. I was hoping something might be a bit more baked re gpt4, but yeah the API isn't publicly available yet and is probably going to cost a pretty penny. If I can make my own model I could in theory run it locally on a timescale that works for my needs though so there is that at least https://www.tensorflow.org/tutorials/images/classification
|
|
#
¿
Apr 16, 2023 23:57
|
|