Not runnable todaySegmentationVerified 2026-11-05
MobileSAM
dhkim2810/MobileSAM (mobile_sam.pt)
MobileSAM is a image segmentation model (Apache-2.0). Not runnable on Mixpeek today.
Something blocks it today, such as the license, remote code or a GPU-only runtime. The reason is below.
No demo for MobileSAM yet
A video, image, slideshow or playground goes here once we record one on our own data.
What it is good and bad at
- Best at
- SAM with a 5M-parameter TinyViT encoder: box or point prompted masks on CPU in tens of milliseconds.
- Worst at
- Mask quality below SAM 2; images only.
- Task
- Segmentation
- Input
- Image
- Output
- Masks, Scores
- Use cases
- Segmentation, Object detection
Templates that use it
No template uses this model yet. Browse templates.
Input and output
Input
- image
- resolution
- 1024x1024
- prompts
- points and boxes
Output
- masks
- binary masks with IoU predictions
How Mixpeek wires it
- Extractors
- None today
- Feature URIs
- None today
- Vector indexes
- None
- Retriever stages
- None
- Combines with
- detector boxes -> MobileSAM masks -> masked crops -> SigLIP
How it runs
- Status
- Not runnable today Weights are a .pt pickle and the model code is the repo's (segment_anything fork). The repo ships an ONNX export script; once exported to ONNX it runs as a CPU plugin with onnxruntime.
- Serve apps
- None
- Runtime
- mobile_sam package (sam_model_registry['vit_t']) or onnxruntime after export
- Weights
- Pickle, no remote code
- Parameters
- 10 M
- Size on disk
- 41 MB
- Preprocessing
- Resize longest side to 1024, pad, normalize.
- Postprocessing
- Threshold mask logits, pick by predicted IoU.
- Serving
- both
- Hardware
- Runs on CPU (slower without a GPU)
Scale to zero
- Minimum replicas
- 0
- Maximum replicas
- 1
- Idle timeout
- 600 s
- Memory
- 1 GB
- CPUs
- 1
- GPUs
- 0
- Cold start, node warm
- Not recorded
- Cold start, new node
- Not recorded
Measured cost
Unmeasured
No committed measurement; plan row 3 measures it in the standalone image.
Our evaluations
Unmeasured
We have not evaluated this model on our own data yet. Publisher results are in the sources below.
Questions
- What is MobileSAM best at?
- SAM with a 5M-parameter TinyViT encoder: box or point prompted masks on CPU in tens of milliseconds.
- What are the limits of MobileSAM?
- Mask quality below SAM 2; images only.
- Can MobileSAM be used commercially?
- Yes. Apache-2.0 per the GitHub repo (the hub mirror is tagged MIT); no use restrictions either way.
- How does MobileSAM run on Mixpeek?
- Weights are a .pt pickle and the model code is the repo's (segment_anything fork). The repo ships an ONNX export script; once exported to ONNX it runs as a CPU plugin with onnxruntime.