AI Visual Recognition for Heavy Duty Aftermarket Inventory
Table of contents
Business Objective
A US heavy duty aftermarket network manages millions of parts across hundreds of categories and manufacturers. Staff rely on manual catalog lookups and institutional knowledge, making counter productivity dependent on veteran expertise.
The network needed a faster, more reliable way to identify unfamiliar parts at scale.
The Solution
OLSYS designed and deployed an AI-powered image recognition platform that replaces manual catalog searches with a brief video clip – eliminating the need for continuous frame streaming, barcode lookups, or part-number reliance.
Counter staff record a short clip on a smartphone and tap to upload. Cloud-based deep learning classifiers evaluate the part’s category; high-confidence outputs pass directly to spatial filtering, while uncertain results are gated to prevent false positives downstream
The platform executes a single-pass query fusing computer vision feature embeddings with physical parameters. Counter staff input FMSI length, width, and a single qualifying attribute, which are filtered against strict tolerance ranges. Surviving candidates are ranked by confidence score, returning exact part numbers and fitment data in seconds.
Results
- High-Precision Visual Identification: Computer vision identifies parts and returns the exact part number and fitment at >98% accuracy.
- Catalog Lookups Eliminated: Manual page-by-page searches are removed entirely, saving ~35 counter staff hours per location monthly.
- Wait Times Cut 60–70%: Parts are identified within seconds, helping new staff serve customers significantly faster.
Let’s Engineer Impact Together
Build software that scales, IT that protects, and AI that delivers. Get in touch to turn your vision into reality.
