Brinq is a comprehensive, vision-based approach for creating real-time analytics at the edge. It is a suite of solutions that combine a microservices architecture, powerful vision algorithms, optimized inference, full-stack integration and edge hardware.
Brinq works entirely at the edge without requiring video to be streamed externally or to a third-party service for analysis. Brinq’s full-stack implementation delivers local real-time situational awareness, notifications, database integration, event timelines, analyst workflows, local actions and configuration.
Brinq makes it easy to build, modify and deploy vision pipelines by combining microservices with runtime orchestration. This architecture facilitates experimentation, reduces code footprint and provides limitless scalability to span pipelines across local compute units all the way to cloud.
Brinq delivers mission critical inference by using industry proven, edge optimized runtimes including TensorFlow Lite, TensorRT and ArmNN with specialized hardware delegates for real-time performance using accelerators. Model conversion, quantization and fine-tuning is available to optimize Pytorch, tflite, ONNX, Caffe or Keras model formats.
Brinq offers a catalogue of specialized models and datasets developed to improve the accuracy of real-world applications. Optimized for people detection, multiple object tracking, occupancy sensing, people counting or crowd analysis, performance and accuracy is proven using industry standard benchmarks such as CrowdHuman or MOTA.
To deliver real-world edge AI applications takes a cross-disciplined team of experts in ML, data science, vision algorithms, optimization techniques and edge systems. Brinq provides the building blocks, while our team of experts will help you realize the rest.
State-of-the-art inference is only the start of intelligence.
Model and runtime enablement using a scalable containerized microservices pipeline architecture
Edge NPU and GPU acceleration using NXP, VeriSilicon, Nvidia and Kinara
RGB and RGBD vision analysis for inspection and analytics
Intrusion, counting, proximity, crowds, face verification; clothing characterization, PPE, age, gender, emotion, action, behavior
Package monitoring and abandoned package detection with owner identification
License Plate Recognition (LPR), vehicle type and make characterization
Industry leading tracking solutions including long-term and multicamera reidentification
Real-time depth and volumetric analysis for road, rail or industrial inspection
Detection zones, boundaries, real-time notifications, timeline visualization and analyst workflows
Arcturus offers scalable, modular edge AI hardware using processors from NXP, NVIDIA and Kinara with accelerated runtime support for TensorFlow Lite, TensorRT and Arm NN.
Atlas is a scalable edge inference platform that uses NPUs and GPUs to support up to 23.3 TOPs of ML acceleration. The platform uses SoCs from NXP and NVIDIA with up to 10x Arm v8 CPU cores, 16GB of SDRAM and up to 1TB of on-board storage. Brinq Edge AI and Vision analytics platform software is available for real-time analysis, SDK level access, component enablement and full-stack solutions.
Learn more about how Brinq can help you.
Whether you want to do-it-yourself or have a project that needs access to our specialized resources, Arcturus offers low-cost development kits or tailored engagements to help you get moving quickly.
Talk to us,
we don’t bite
