MAKRR
Turns any existing camera — a security cam, a drone, a robot's eye — into a smart sensor that can spot and count objects, without anyone on the team needing to know machine learning.
🔗 Visit MAKRR
Description
Building a computer vision system traditionally means hiring or becoming a machine-learning engineer: collecting images, labeling them, training a model, then figuring out how to run it fast enough on real hardware. MAKRR compresses that whole process into a point-and-click workflow — you upload video from a camera you already have, mark what you want it to recognize, and it trains and deploys a working detector for you.
MAKRR is a no-code computer vision platform aimed at industrial operations teams and technology vendors: it ingests footage from CCTV, drone, or robot cameras, lets you annotate objects or behaviors, trains a model from a library of 17 production architectures, and deploys it to edge hardware (including bundled Jetson devices on higher tiers) for real-time monitoring — claiming production-ready results in about a day, with 100% on-premise deployment so footage never has to leave the site.
💬 Our review
The short version: if you have existing cameras and a repetitive visual-inspection task (spot a defect, count units, detect a specific event) but no ML team, MAKRR's no-code pipeline gets you there without hiring anyone — the tradeoff is a genuinely industrial price tag, starting at €99/month for a fairly limited 2,000 detections/day.
Roboflow is the closest comparison but leans more toward developers who are comfortable writing some code around the model; MAKRR positions itself further toward zero-code for operations staff. Amazon Lookout for Vision is AWS-native and usage-billed rather than a flat monthly fee, which suits variable workloads better but ties you into AWS infrastructure. MAKRR's on-premise deployment is the standout for anyone who can't send camera footage to the cloud for compliance reasons — that's a real differentiator, not just a marketing line — but it does mean budgeting for edge hardware (Jetson devices) on top of the subscription once you're past the entry tier.
📊 Global score
🤖 AI-enriched data
START 99€/mois (2000 détections/jour, 1 formation/mois). BUILD 300€/mois (2000 détections/jour, 2 formations/mois, 2 appareils edge). OPERATE 750€/mois (10000 détections/jour, 10 formations/mois, 10 appareils Jetson inclus). ENTERPRISE sur devis. Essai gratuit disponible.
Pros
Plateforme no-code, aucune expertise ML nécessaire
Inférence edge haute performance (jusqu'à 260 fps)
Déploiement 100% sur site, données privées
Production en ~1 jour avec 17 architectures de modèles
Cons
Quotas de détections limités sur les plans d'entrée
Coûts additionnels pour appareils edge Jetson
Pas de plan gratuit permanent, essai seulement