SFW API vs AWS Rekognition Content Moderation
Last updated Mon Nov 09 2026 00:00:00 GMT+0000 (Coordinated Universal Time)
Rekognition returns a hierarchy of moderation labels (Explicit Nudity, Suggestive, Violence, …) with confidence values; the SFW API returns one nudity level and a policy verdict.
| SFW API | AWS Rekognition | |
|---|---|---|
| Entry price | $10 for 10,000 credits ($1.00/1k) | about $1.00 per 1,000 images, tiered down with volume |
| Free | 500 credits/month | free tier for 12 months |
| Setup | API key, one request | AWS account, IAM, SDK, region |
| Output | level 0–3, scores, flags, blocked | label tree with MinConfidence, parent/child names |
| Video | frames in the same call | asynchronous video jobs via SNS/SQS |
| Data handling | nothing stored | per AWS data policy; opt-out of improvement use needed |
Choose Rekognition when you are already deep in AWS and want many label types with regional residency choices. Choose the SFW API for a simpler contract, an ordinal level instead of a label forest, synchronous GIF/video, and prepaid pricing.
Mapping labels
| Rekognition label | SFW API |
|---|---|
| Explicit Nudity → Exposed Male/Female Genitalia, Nudity, Sexual Activity | level 3 Explicit |
| Explicit Nudity → Partial Nudity, Suggestive → Revealing Clothes (with little coverage) | level 2 Implied Nudity |
| Suggestive → Female Swimwear Or Underwear, Male Swimwear Or Underwear, Revealing Clothes | level 1 Revealing |
| Sex Toys | sex_object flag |
| Illustrated Explicit Nudity | art flag + level |
MinConfidence (0–100) maps roughly to our scores: a level counts above 50 on a calibrated scale, so start with 50 rather than Rekognition's usual 80.