AI-Powered Quality Control for Precision Brass Inserts & Fasteners
Project description
The AI Sorting Machine is a 3-station automated visual inspection system that uses deep learning and classical computer vision to detect 12 categories of defects on precision brass inserts, screws, and fasteners — replacing inconsistent manual inspection with sub-100ms per-part AI inference.
Each part is transported through three inspection stations, rotating 360° on its own axis at each point. Station A uses a Basler racer 2 line scan camera with a telecentric lens to capture a full unwrapped surface image, running YOLOv8-Seg instance segmentation to detect missing or damaged knurling patterns and surface defects. Station B uses a Basler ace area scan camera with a bore inspection lens to examine internal threads, detecting missing threads, absent chamfers, burrs, and foreign particles through an EfficientNet-B3 classifier combined with PatchCore anomaly detection. Station C uses a Mind Vision camera with telecentric backlit silhouette imaging for dimensional verification — measuring part height, diameter, slot width, and across-flats tolerance to sub-millimetre precision, supplemented by a ResNet-18 classifier for stamp presence detection.
The system targets a false pass rate below 0.5%, meaning fewer than 5 defective parts per 1,000 escape detection — a standard that significantly exceeds what human inspectors can sustain across full production shifts. All inference runs on ONNX Runtime / TensorRT with GPU acceleration, and results are logged with full image traceability for quality audit.
What We Are Seeking:
We are seeking technology integration partners with expertise in industrial machine vision deployment who can support the transition from prototype to production-grade system — particularly in line scan image acquisition pipeline optimization, PLC integration for reject actuation, and on-site commissioning at Bezalel's manufacturing facility. We are also seeking follow-on projects from fastener and automotive component manufacturers in South Asia who face similar quality control challenges and would benefit from a sovereign, AI-driven visual inspection platform.
Why this Matters Now
Manual visual inspection remains the dominant quality control method across South Asia's fastener manufacturing sector, employing human inspectors who work in shifts examining thousands of small brass parts per day. This approach is fundamentally limited: inspector fatigue causes error rates to climb after the first few hours, subtle defects like partial knurling damage or shallow burrs inside threads are routinely missed, and the labour cost of multi-shift inspection teams is significant. As automotive and electronics OEMs tighten incoming quality requirements — particularly under IATF 16949 and zero-defect supply chain mandates — manufacturers who cannot demonstrate automated, traceable inspection will lose contracts to competitors who can. The window for establishing AI-driven inspection capability in this market is open now, before larger automation vendors move downstream from semiconductor and pharmaceutical inspection into general fastener QA.
Why Bhutan and Why InnoTech
DHI InnoTech's advantage lies in its position as Bhutan's applied AI development centre with direct access to low-cost, hydropower-backed compute infrastructure and a growing team of AI/ML engineers already delivering production systems (including the Bhutan Sovereign ANPR system for the Royal Bhutan Police). The DRIVE Division combines software-defined AI capability with hands-on mechanical and electrical integration — the team building this system includes specialists in computer vision model training, embedded systems, PLC integration, 3D-printed fixture design, and industrial camera calibration, all operating under one roof. This integrated capability allows InnoTech to deliver a complete vision system at a fraction of the cost of European or Japanese machine vision integrators, while maintaining full sovereignty over the AI models, training data, and inference pipeline — a critical differentiator for clients concerned about IP and data control.
Partnerships and Collaboration
The project operates as a direct B2B engagement between DHI InnoTech and Bezalel Skilligence, with Bezalel providing sample parts, defect specifications, and production environment access, while InnoTech delivers the complete AI vision system. The development team comprises 7 members across mechanical design, electrical integration, and AI/software engineering, with clearly defined station-level ownership.
Want to collaborate on this project? Write the lead directly, or pitch an adjacent idea through the open queue.
Our objective is to establish a national-scale, technology-enabled GLOF risk monitoring and mitigation platform, beginning with high-risk catchments such as the Punatsangchhu basin.
An IoT-enabled Smart Water Management System designed to improve water distribution, reduce wastage, and ensure reliable access to safe drinking water. The system integrates smart sensors for water quality, pressure, and flow monitoring, supported by remote monitoring and control to detect anomalies and optimize water supply from source to end users. Currently being piloted in Bhutan, the solution has potential for regional and international scale-up.
A mini ‘Fab Lab’ platform that equips Bhutanese schools with digital fabrication tools, teacher training, and ongoing technical support to drive hands-on STEAM learning and student-led innovation.
The proposed geothermal program is a multi-phase initiative led by Druk Holding and Investments (DHI) to unlock Bhutan’s geothermal potential across multiple Dzongkhags such as Gasa, Punakha, and Gelephu:
Electricity Generation – Deploy small-scale electricity generation providing baseload renewable power to the grid.
Druk Holding and Investments (DHI) was established with a mandate to enhance the performance and value of its portfolio companies and safeguard national wealth for all generations of Bhutanese through prudent investments. Over the past decade, DHI has built a diversified group of companies operating across key sectors — energy, telecommunications, finance, manufacturing, technology, and services.
