Meta Segment Anything Model 3

(ai.meta.com)

177 points | by alcinos 6 days ago

13 comments

  • trevorhlynn 8 hours ago
    This was front page for a while last week

    https://news.ycombinator.com/item?id=45982073

  • vessenes 6 hours ago
    Released last week. Looks like all the weights are now out and published. Don’t sleep on the SAM 3D series — it’s seriously impressive. They have a human pose model which actually rigs and keeps multiple humans in a scene with objects, all from one 2D photo (!), and their straight object 3D model is by far the best I’ve played with - it got a really very good lamp with translucency and woven gems in usable shape in under 15 seconds.
    • Qwuke 5 hours ago
      Between this and DINOv3, Meta is doing a lot for the SOTA even if Llama 4 came up short compared to the Chinese models.
    • nl 5 hours ago
      https://ai.meta.com/blog/sam-3d/ for those interested.
    • Fraterkes 5 hours ago
      Are those the actual wireframes they're showing in the demos on that page? As in, do the produced models have "normal" topology? Or are they still just kinda blobby with a ton of polygons
      • vessenes 1 hour ago
        I’ve only used the playground. But I think they are actual meshes - they don’t have any of the weird splat noise at the edge of the objects, and they do not seem to show similar lighting artifacts to a typical splat rendering.
      • seanw265 3 hours ago
        I haven’t tried it myself, but if you’re asking specifically about the human models, the article says they’re not generating raw meshes from scratch. They extract the skeleton, shape, and pose from the input and feed that into their HMR system [0], which is a parametric human model with clean topology.

        So the human results should have a clean mesh. But that’s separate from whatever pipeline they use for non-human objects.

        [0]: https://github.com/facebookresearch/MHR

      • daemonologist 3 hours ago
        For the objects I believe they're displaying Gaussian splats in the demo, but the model itself can also produce a proper mesh. The human poses are meshes (it's posing and adjusting a pre-defined parametric model).
    • visioninmyblood 3 hours ago
    • retinaros 2 hours ago
      I looked quickly but it does not generate a 3d model file right?
  • enoch2090 5 hours ago
    Surprisingly, SAM3 works bad on engineering drawings while SAM2 kinda works, and VLMs like Qwen3-VL works as well
    • zubiaur 2 hours ago
      Had good luck with Gemini 2.5, SAM3 failed miserably with PIDs.
    • retinaros 2 hours ago
      yeah I tried too. Im trying a fine tuning on PIDs.
  • vanjoe 2 hours ago
    For a long time I've wanted to use something like this to remove advertisements from hockey games.The moving ads on the boards are really annoying. Maybe I'll get around to actually doing that one of these days.
  • the_duke 7 hours ago
    Side question: what are the current top goto open models for image captioning and building image embeddings dbs, with somewhat reasonable hardware requirements?
    • daemonologist 2 hours ago
      For pure image embedding, I find DINOv3 to be quite good. For multimodal embedding, maybe RzenEmbed. For captioning I would use a regular multimodal LLM, Qwen 3 or Gemma 3 or something, if your compute budget allows.
    • NitpickLawyer 7 hours ago
      Try any of the qwen3-vl models. They have 8, 4 and 2B models in this family.
    • Glemkloksdjf 7 hours ago
      I would suggest YOLO. Depending on your domain, you might also finetune these models. Its relativly easy as they are not big LLMs but either image classification or bounding boxes.

      I would recommend bounding boxes.

      • jabron 5 hours ago
        What do you mean "bounding boxes"? They were talking about captions and embeddings, so a vision language model is required.
        • Glemkloksdjf 1 hour ago
          I suggested YOLO and non llm-vl as a lot faster alternative.

          Of course CLIP would be otherwise the other option than a big llm-vl one.

      • smallerize 6 hours ago
        Which YOLO?
        • Glemkloksdjf 6 hours ago
          Any current one. they are easy to use and you can just benchmark them yourself.

          I'm using small and medum.

          Also the code for using it is very short and easy to use. You can also use ChatGPT to generate small exepriments to see what fits your case better

          • throwaway314155 5 hours ago
            There aren’t any YOLO models for captioning and the other models aren’t robust enough to make for good embedding models.
            • Glemkloksdjf 1 hour ago
              You can get labels out of the classifier and bounding box models.

              They are super fast.

              Its just an alternative i'm mentioning. I would assume a person knowing a little bit of that domain.

              Otherwise the first option would be CLIP i assume. llm-vl is just super slow and compute intensive.

  • phkahler 5 hours ago
    Which (if any) of these models could run on a RaspberryPi for object recognition at several FPS?
  • aliljet 4 hours ago
    I wonder how effective this is medical scenarios? Segmenting organs and tumors in cat scans or MRIs?
  • maelito 3 hours ago
    I wonder if this can be used to track an object's speed. E.g. a vehicle on a road. It would need to recognize shapes, e.g. car model or average size of a bike, to guess a speed.
  • colkassad 5 hours ago
    Been waiting days to get approval to download this from huggingface. What's up with that?
    • observationist 3 hours ago
      Alternative downloads exist. You can find torrents, and match checksums against the HF downloads, but there are also mirrors and clones right there in HF which you can download without even having to log in.
    • knicholes 3 hours ago
      I was approved within about 10 minutes for "Segment Anything 3"
    • tschellenbach 3 hours ago
      same here, didn't get approval
  • shashanoid 5 hours ago
    Miss the old segment anything page, used it a lot. This UI I found very complex to use
  • cheesecompiler 5 hours ago
    This would be convenient for post-production and editing of video, e.g. to aid colour grading in Davinci Resolve. Currently a lot of manual labour goes into tracking and hand-masking in grading.
  • Will-Reppeto 5 hours ago
    [dead]
  • Workaccount2 8 hours ago
    I do a test on multimodal LLMs where I show them a dog with 5 legs, and ask them to count how many legs the dog has. So far none of them can do it. They all say "4 legs".

    Segment anything however was able to segment all 5 dog legs when prompted to. Which means that meta is doing something else under the hood here, and may lend itself to a very powerful future LLM.

    Right now some of the biggest complaints people have with LLMs stems from their incompetence processing visual data. Maybe meta is onto something here.