Overview
The COVID-19 epidemic in early 2020 became a worldwide pandemic halfway through 2020. Being a lower respiratory infectious virus, an effective way to minimize the chances of contracting the virus was by wearing a mask. Suddenly, mask manufacturers were faced with demands that greatly exceeded their standard manufacturing capacity. The general medical-grade mask manufacturing process involves three layers of cloth cut into specific sizes (for adults and children use), laid out and folded and straps are stamped/ stuck or stitched onto the four corners. Thereafter, each mask is inspected for cloth/ fabric stain or tears, faulty strap stitching or attachment by factory personnel.
Challenges
Due to the pandemic outbreak, demands for medical-grade disposal masks rose sharply. While the automated production was swift, unfortunately, the back-end manual labor inspection for faults could not meet the hundreds of millions needed for distribution. As the quality-checking process is an important part of the whole disposal medical-grade mask production, an effective way to fault-check the produced masks and ramp up the inspection process was to integrate automated inspection systems that can quickly learn and distinguish minor differences or faults on the production line. This can guarantee the quality of the mask’s production process while dramatically increasing the quantity produced daily.
Solution
Neousys Edge AI GPU Platform
To address these issues, our customers chose Neousys Technology controllers that feature machine learning AI inference capabilities coupled with assisted-lighting and cameras. The Neousys edge AI GPU platforms can utilize a sample size as small as 30 to 50 as references for the system to quickly learn the faults it needs to look and scan for. Faults such as stains, rips, faulty strap attachment, the width of the mask, etc. can all be scanned efficiently and automatically to dramatically multiply the daily production volume.
Neousys edge AI GPU platforms range from power-efficient Jetson™ Xavier, mainstream cost-effective RTX 30 series graphics cards to high performance and demanding Tesla/ Quadro inference accelerators. Systems can come with up to eight PoE+ connectivity for ease of connecting (single cable for power and data) to the mask scanning cameras. The detected faulty masks are tagged in the lineup and scheduled to be re-inspected or removed at the designated station along the production line.
Benefits of Neousys Edge AI GPU Platform
Neousys ruggedized edge AI GPU platforms offer the following advantages:
- IP67 water/ dustproof GPU computer (model specific)
- Extreme rugged computers for extra volatile environmental conditions (model specific)
- Shock and vibration proof for in-vehicle use
- True -25°C to 70°C wide-temperature operations for harsh, volatile environments
- Support up to NVIDIA® RTX 30 series graphics cards (model specific)
- Support up to NVIDIA® Tesla/ Quadro inference accelerators (model specific)
- PoE+ connectivity for GigE cameras
- USB3.1 Gen1/ Gen2 connectivity for USB cameras
- WiFi/ 4G/ 5G expansion via mini-PCIe module
- Patented Cassette design for segregating thermal/ electrical interference
- Patented SuperCAP UPS to counteract unforeseen power interruptions
