NEWVectors or files. Pick a path.Start →
    Served on MixpeekClassificationRecommendedVerified 2026-11-05

    Falconsai NSFW image detection (ViT)

    Falconsai/nsfw_image_detection

    Falconsai NSFW image detection (ViT) is a image and video classification model (Apache-2.0). Served on Mixpeek, scaling to zero when idle.

    A Ray Serve app in Mixpeek's inference registry hosts it. It starts at zero replicas and wakes on first use.

    Use in a new namespaceAdd to an existing namespace Record as JSON

    New opens the Designer in Studio. For an existing namespace, open it in Studio and use its Designer tab.

    No demo for Falconsai NSFW image detection (ViT) 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
    Fast binary normal/nsfw screening of images, and of video by the max score over sparse sampled frames.
    Worst at
    Graded or policy-specific safety (violence, drugs, suggestive but allowed content): it has two labels only; one bad frame between samples can be missed.
    Task
    Classification
    Input
    Image, Video
    Output
    Labels, Scores
    Use cases
    Moderation, Classification
    Why we recommend it
    Served CPU-only image safety classifier behind the content classifier app.

    Templates that use it

    No template uses this model yet. Browse templates.

    Input and output

    Input

    image
    resolution
    224x224 (ViTImageProcessor)
    color
    RGB
    video
    sparse frames sampled by engine.inference.frame_sampling

    Output

    labels
    normal, nsfw
    scores
    nsfw_score in [0, 1]; is_nsfw = nsfw_score >= threshold (default 0.5)

    How Mixpeek wires it

    Extractors
    None today
    Feature URIs
    None today
    Vector indexes
    None
    Retriever stages
    classify
    Combines with
    • classify task=nsfw calls mixpeek/content-classifier-v1 per document (image or video URL); text documents go to the DistilBERT text classifier in the same app

    How it runs

    Status
    Served on Mixpeek Image and video half of the mixpeek__content_classifier_v1 Serve app.
    Serve apps
    mixpeek__content_classifier_v1
    Runtime
    transformers pipeline("image-classification") (ViTForImageClassification)
    Weights
    safetensors, no remote code
    Parameters
    85.8 M
    Size on disk
    343 MB
    Preprocessing
    Flatten to RGB, pipeline image processor (resize 224, normalize); video sampled to sparse frames.
    Postprocessing
    Softmax over both labels (top_k=None); nsfw_score is the max over labels matching nsfw/unsafe/porn/explicit/questionable, max over frames for video.
    Serving
    realtime
    Hardware
    Runs on CPU (slower without a GPU)

    Scale to zero

    Minimum replicas
    0
    Maximum replicas
    4
    Idle timeout
    300 s
    Memory
    2 GB
    CPUs
    2
    GPUs
    0
    Cold start, node warm
    Not recorded
    Cold start, new node
    Not recorded

    Measured cost

    Unmeasured

    No committed throughput or cost measurement for this model.

    Our evaluations

    Unmeasured

    We have not evaluated this model on our own data yet. Publisher results are in the sources below.

    Questions

    What is Falconsai NSFW image detection (ViT) best at?
    Fast binary normal/nsfw screening of images, and of video by the max score over sparse sampled frames.
    What are the limits of Falconsai NSFW image detection (ViT)?
    Graded or policy-specific safety (violence, drugs, suggestive but allowed content): it has two labels only; one bad frame between samples can be missed.
    Can Falconsai NSFW image detection (ViT) be used commercially?
    Yes. Apache-2.0 on the repo; a community fine-tune of a ViT base, with no published training-data audit.
    How does Falconsai NSFW image detection (ViT) run on Mixpeek?
    Image and video half of the mixpeek__content_classifier_v1 Serve app.