[!] BETA · v0.9API, model and pricing may change before 1.0 · talk to us before shipping commercial production → support@antinude.io
ANTINUDE.IO / SDK v0.9 / iOS · ANDROIDALL SYSTEMS · 47ms p50
Nudity detectionImage + videoOn-device
Content safety,
on-device.
50 ms or less.
Drop in our SDK. Detect nudity in images and video — keyframe-sampled, on device, in under 50 ms — with a confidence score you can act on.
99.94%
precision
on held-out test set
47 ms
median latency
iPhone 12 · iOS 17
100k+
training images
12 cultural contexts
12 MB
SDK size
incl. model weights
Swift · iOS · 3 linescopy
1import ANSdk
2
3let client = try ANClient(apiKey: "ak_live_…")
4let result = try await client.scanImage(data)
5
6// → result.verdict == "safe"
7// → result.topScore == 0.02
Kotlin · Android · 3 linescopy
1import com.an.sdk.ANClient
2
3val client = ANClient.create(ctx, apiKey = "ak_live_…")
4val result = client.scanImage(bytes)
5
6// → result.verdict == "safe"
7// → result.topScore == 0.02
§01 / PIPELINEHow it runs · on-device.
STEP 01
User image / video
camera · gallery · upload
keyframes sampled @ 1 fps
>>
STEP 02
AntiNude SDK
CoreML / TFLite
47 ms · 12 MB · on-device
>>
STEP 03
Your verdict
"safe" / "unsafe"
+ per-class detections
→ image & video bytes never leave the device · cloud is only used for license check
§02 / COMPLIANCESafety & privacy by default.
[✓]On-device inference
[✓]Image & video keyframes
[✓]Bytes never leave device
[✓]Encrypted in transit & at rest