Pixicular is a B2B photo analysis API. You send us an image (file upload or URL), pick which services you want, and get structured JSON back. Currently live with five services:
• detect-age — age estimation of people in the image
• detect-labels — general object and scene label detection (e.g. “Outdoor”, “Dog”, “Beach”)
• detect-face-emotions — face emotion detection and analysis
• detect-moderation — moderation labels (nudity, violence, etc.)
• detect-text — text detection and extraction from images
The API runs on Fastify/Node.js on a Hetzner server in Nuremberg, with Amazon Rekognition doing the heavy lifting on the ML side. The frontend is Next.js on Vercel. Infrastructure is Terraform on AWS (S3, CloudFront with signed URLs, IAM Identity Center). PostgreSQL for data, Clerk for auth, Stripe for billing.
I’m also training a custom AI-generated image detection model (hosted on Hugging Face), but it’s not accurate enough for production yet, so I shipped these five Rekognition-backed services first rather than waiting. The AI detection service will be added once the model is ready.
The Developer tier is free — 100 checks/month, no credit card required. Paid plans start at $25/mo for 1,000 checks.
Would love to hear feedback on the API design, pricing, or use cases you’d find valuable. What other detection service(s) would you like to have added that would bring the best value? Happy to answer any questions about the architecture, too.
Wow, I didn't know Gentoo is still around.
I was on Gentoo for a while back in 2001-2002 or so. Wonderful distro. What a treat after Red Hat Linux 6.2. Then, a few years later, I switched to Linux Mint etc. Nowadays, I'm on macOS, but when I have a bit more time, I will give Gentoo a go again.
Pixicular is a B2B photo analysis API. You send us an image (file upload or URL), pick which services you want, and get structured JSON back. Currently live with five services:
• detect-age — age estimation of people in the image • detect-labels — general object and scene label detection (e.g. “Outdoor”, “Dog”, “Beach”) • detect-face-emotions — face emotion detection and analysis • detect-moderation — moderation labels (nudity, violence, etc.) • detect-text — text detection and extraction from images
The API runs on Fastify/Node.js on a Hetzner server in Nuremberg, with Amazon Rekognition doing the heavy lifting on the ML side. The frontend is Next.js on Vercel. Infrastructure is Terraform on AWS (S3, CloudFront with signed URLs, IAM Identity Center). PostgreSQL for data, Clerk for auth, Stripe for billing.
I’m also training a custom AI-generated image detection model (hosted on Hugging Face), but it’s not accurate enough for production yet, so I shipped these five Rekognition-backed services first rather than waiting. The AI detection service will be added once the model is ready.
The Developer tier is free — 100 checks/month, no credit card required. Paid plans start at $25/mo for 1,000 checks.
Would love to hear feedback on the API design, pricing, or use cases you’d find valuable. What other detection service(s) would you like to have added that would bring the best value? Happy to answer any questions about the architecture, too.
Cheers!
Martin