What industrial manufacturing challenge does the V-Deelia Deep Learning Imaging Analytics solution resolve?

Traditional SMT machine vision setups rely on rigid, rule-based algorithmic thresholds that frequently generate excessive false call rates (FCR) due to minor variations in component cosmetics or background reflectivity. This pseudo-defect volume overloads manual buy-off stations, forcing operators to inspect massive image streams. Over long shifts, operator visual fatigue peaks, causing critical defect escapees. V-Deelia resolves this by deploying self-learning neural networks that automatically filter cosmetic variations from true functional defects.

How does V-Deelia automate defect classification to optimize verification station labor metrics?

The software incorporates advanced Automated Defect Classification (ADC) models trained on vast manufacturing image repositories. When inline hardware flags an anomaly, the V-Deelia processing engine evaluates the image features using deep neural networks, instantly reclassifying benign variations without manual intervention. This shifts human attention exclusively to true component failures, cutting down labor verification overhead significantly.

Can the V-Deelia software engine catch micro-defects that traditional rule-based programming misses?

Yes. While traditional pixel-matching vision tools struggle with low-contrast, shifting grayscales, or complex component layouts, V-Deelia utilizes abstract feature extraction via deep learning. This allows the system to consistently isolate micro-defects—such as minor substrate contamination, early component micro-cracks, and micro-chipping—even under challenging illumination variations.

How does this deep-learning image analytics suite integrate with the overarching V-ONE ecosystem?

V-Deelia operates as an intelligent core analytics engine embedded straight within the V-ONE Smart Manufacturing dashboard architecture. The data classification results are cross-referenced with centralized machine health metrics, tracking live false call trends, true defect distributions, and yield optimization data to orchestrate proactive adjustments across distinct SMT process zones.

What hardware line configurations and electronic device families are supported by the V-Deelia AI platform?

Because V-Deelia's deep learning algorithms focus on generalized structural and pixel anomalies rather than rigid component footprints, it adapts to diverse electronic hardware families. The platform provides automated quality optimization for high-density consumer electronics assemblies, rugged automotive powertrain modules, high-reliability medical computing blocks, and intricate backend semiconductor devices.

V-ONE V-Deelia Deep Learning Imaging Analytics

Looking to improve inspection accuracy and efficiency? Our cutting-edge technology integrates with any device and harnesses the power of A.I., reducing inspection times and increasing accuracy.

Let us introduce you to DeeLIA, the auxiliary tool designed for vision inspection processes – read on to learn more!

DeeLIA, also known as Deep Learning Imaging Analytics, is a powerful tool that enhances automated visual inspection of manufactured products. It adopts advanced algorithms and artificial intelligence technology to elevate inspection performance, delivering a remarkable decrease in false call rate while maximising yield. The available models in DeeLIA are the:

ViTrox Patented V-Model
Classification
Object Detection, Segmentation and Object Attention (patent pending).

At ViTrox, our comprehensive approach to getting started with DeeLIA involves a six-stage process (as shown below). Firstly, we begin with data collection, extracting both good and bad/rejected images from the database. Secondly, we move on to data cleaning, to ensure clear labelling of defect images into classes to identify and categorise the defect types accurately. The third stage then focuses on data validation, where we meticulously verify the collected data within each class, ensuring its proper placement aligns with the quality department’s requirements. Following this, the fourth stage involves thorough data labelling, where we provide detailed categorisation of images to facilitate deep learning model training. The fifth stage is dedicated to model training, where we select the appropriate model and train it using the validated and labelled data. Lastly, the sixth stage entails model validation, where we confirm the model's effectiveness by applying it to new production data and evaluating its performance in predicting outcomes. Through this comprehensive process, ViTrox leverages the power of artificial intelligence to deliver accurate and efficient imaging analytics solutions with DeeLIA.

DeeLIA: Your Inspection Breakthrough

DeeLIA improves the manufacturing process by ensuring product quality, improving speed and efficiency, and reducing costs. It's a solution that provides many benefits for the manufacturing industry. Here are three major reasons why you need DeeLIA:

Support Different Inspection Devices
DeeLIA offers an extensive range of defect detection capabilities for different industries. With DeeLIA's existing model, it can perform Optical Character Recognition/Verification, Object Parts Detection, and Foreign Material Detection. Its coverage includes:
- SMT: Polarity Detection, Object Skewed Detection, Object Offset Detection, Flip, Tombstone, Billboard, Bridging, and Solder Insufficiency/Excess.
- Semiconductor: Void (Fish Scales), Filament, Peeling, Scratches, and Edge Cracking.
- Others: Cracks and Bubbles.
Reduce Overkill and False Call Rates while Minimising Human Error
False calls and overkill are common issues that happen in the production line. Thus, it often requires human judgement to determine the final result, which can be prone to errors too. Now with DeeLIA, images flagged as false calls by inspection machines will be reevaluated to assess whether it is a “true defect” or a “false call”, which effectively saves and reduces the production of unnecessary items. With the reliable inspection models of DeeLIA, users can have complete confidence in the precision and accuracy of the quality assessment. This function ensures that your products meet high standards while saving time and cutting costs.
Remains Compatible with Your Current Inspection Devices
DeeLIA can be integrated into your existing inspection system/devices with the help of additional edge AI computing. The edge AI computing will contain the trained model inference into it and be processed at lightning speed with the help of a graphical processing unit (GPU). Plus, you gain better insights into your current inspection processes, which helps you identify areas of improvement.

Key Takeaway

In conclusion, DeeLIA propels inspection technology to new heights, delivering substantial enhancements in process efficiency, product quality, and cost-effectiveness, truly surpassing expectations. If you want to learn more about how DeeLIA can take your manufacturing operations to the next level, reach out to us at enquiry@vitrox.com, and our team will be happy to provide you with more information.