MAKRR
MAKRR provides a no‑code computer‑vision platform that lets users upload video recordings or connect live feeds from CCTV, RTSP, USB cameras, drones or robots, tag objects to recognize, and generate a deployable edge AI model in under a day. The service automates training and inference, delivering real‑time insights such as person or vehicle detection without requiring any coding expertise.
- Artificial Intelligence
- Industrial Automation
Funding
Founders
Product
Problem
Many organizations rely on manual visual inspection of video feeds to detect objects, defects, or safety issues, requiring specialized data‑science expertise and cloud‑based processing that can be slow, costly, and raise data‑privacy concerns.
Solution
Makrr provides a no‑code computer‑vision platform that lets users upload recorded video or connect live feeds from existing CCTV, RTSP, USB cameras, drones, or robots. Users tag the objects or behaviors they want to detect, set a priority, and with a single click the system trains an AI model and deploys it to on‑premise edge devices in under a day. The platform runs inference locally, so footage never leaves the user’s network, preserving privacy and reducing latency. Real‑time insights such as vehicle, person, or defect detection are delivered without any coding or a dedicated ML team, enabling automated sorting, workforce optimization, and other operational improvements.
Target Audience
Primary customers are manufacturers, healthcare facilities, logistics operators, and security teams that need rapid, on‑site visual inspection and automation using existing camera infrastructure.
Features
- Drag‑and‑drop interface for uploading video or linking live camera streams (CCTV, RTSP, USB, drones, robots)
- One‑click labeling and priority setting that triggers automated model training without writing code
- Edge deployment of inference models that run on‑premise, eliminating cloud latency and keeping data on the user’s network
- Real‑time object detection with pre‑trained classes (car, truck, bus, van, person, trailer, motorcycle) and custom class support
- Fast model turnaround: footage to deployed model in under a day, with inference speeds up to 260 fps
- No‑code workflow eliminates the need for a data‑science team, allowing domain experts to teach the system directly