Thanks for your submission

We’ll get in touch shorty

Oops! Something went wrong while submitting the form.

Thanks for your submission

We’ll get in touch shorty.

Oops! Something went wrong while submitting the form.

The file is downloading

Please check in a nearby tab.

Or download here

Oops! Something went wrong while submitting the form.

The file is downloading

Please check in a nearby tab.

Or download here

Oops! Something went wrong while submitting the form.

The file is downloading

Please check in a nearby tab.

Or download here

Oops! Something went wrong while submitting the form.

The file is downloading

Please check in a nearby tab.

Or download here

Oops! Something went wrong while submitting the form.

For Logistics and Manufacturing

Deliver Top Performing Computer Vision in Hours

Leverage Generative AI and Synthetic Data to increase efficiency and reduce cost.

Oops! Something went wrong while submitting the form.

Vision

Vision AI Needs a lot of data but no one said it had to come from real photos...

Data Generation

Faster time to value

Reduce data collection time and cost by creating thousands of synthetic images from only 5-10 real images.

Dataset Improvement

Higher accuracy

Improve the quality of your dataset by finding and creating the images you are missing for a high performing model.

Instant Adaptability

Reduce downtime

Adapt quickly to new products, defects, and process changes by generating synthetic data and retraining your model on the fly.

Feature highlight

Product demo

Discover the key features and benefits that make our solution stand out.

Proof of efficiency

Case studies

These stories showcase real achievements of our clients and demonstrate the effectiveness of our solutions in practical scenarios.

Advex Case Study:
Accurate Defect Detection for Automotive Components

PROBLEM: Limited Data for Defect Detection

A major auto manufacturer achieved only a .3 IoU for defect detection on small components because images of defective parts were difficult to collect.

Sample synthetic images reflecting real defects

SOLUTION: Synthetic Images of Rare Defects

Advex generated over 10k high-quality synthetic images to supplement the existing dataset.

With Advex, detecting small but important defects is possible.

RESULT: 52% Fewer Defects Escaped

Advex Case Study:
MVTec Carpet Defect Detection

PROBLEM: Unable to detect all defects on carpets

Automotive supplier could not detect all defective products on a production line.

Based on Real (left), Synthetics (right) can have multiple and more extreme defects

SOLUTION: Synthetic Data for Rare Defects

Advex automatically doubled the existing dataset by generating high quality synthetic images.

With Advex, more defects are found.

RESULT: 5x better defect detection

Advex Case Study:
Accurate Automated Truck Unloading

PROBLEM: Unable to detect all boxes accurately

Robotic Truck Loading/Unloading company could not detect all boxes, especially damaged and dimly lit boxes.

Advex Generated Synthetic Data for Truck Unloading

SOLUTION: Synthetic Data for Difficult Cases

Advex automatically found gaps in customer’s dataset and generated 10k+ high-quality synthetic and labeled images.

With Advex, all boxes are found and there are no erroneous bounding boxes.

RESULT: 75% fewer missed boxes in 1 week

Advex Case Study:
Synthetic Data Reflecting New Site

PROBLEM: Long setup times for each new deployment

Fortune 500 company’s Automated Robotic Truck Unloading system took 2 weeks to fine-tune for every new customer site.

Advex Generated Synthetic Data for Truck Unloading

SOLUTION: Synthetic Data Reflecting New Site

With only 10 images from each new site, Advex generated 2k synthetic images in just 12 hours

With Advex, all boxes are found and there are no erroneous bounding boxes.

RESULT: Customer supports new sites 14x faster

The company was able to clear a back log of customer deployments, reducing deal time and freeing resources for growth.

Advex Case Study:
Accurate Automated Truck Unloading

PROBLEM: Can’t Handle Diverse Packages / Boxes

Fortune 500 logistics firm could only achieve an F1 accuracy of 55% on diverse customer parcels despite having 10,000 images.

Synthetic data of diverse parcels on various conveyor belts

SOLUTION: Synthetic Data for Difficult Cases

Advex automatically identified key gaps in customer dataset and generated right volume of synthetic images for each gap

33% improvement in F1 Accuracy

RESULT: 33% Higher F1 Accuracy in far less time

Advex Case Study:
Weather Variation for Self Driving

PROBLEM: Unable to operate in all weather conditions

Self Driving Car Company could not easily collect data in all weather conditions leading to poor performance in non-sunny conditions.

Based on a real image (left), Advex generates synthetics (right)

SOLUTION: Synthetic Data for Weather Variations

Advex automatically generated synthetic data of snow, fog, rain, and sleet to balance the dataset.

Example Synthetic Images

RESULT: Reach Production Years Faster

Advex Case Study:
Accurate Weed Detection for Autonomous Weeder

PROBLEM: Unable to detect new weeds

Autonomous weeding robot company could not accurately add new weed species without significant data collection and fine-tuning.

Detection Before Advex (Left) and After Advex (Right)

SOLUTION: Synthetic Data For Weeds

With just 10 real images of new weeds, Advex was able to create thousands of synthetic, labeled images.

Advex Generated Synthetic Data

RESULT: Customer supports new farms 10x faster

With Advex, the company can support new customer farms in 2 days instead of 4 weeks.

Advex Case Study:
Accurate Defect Detection for Produce

PROBLEM: Unable to detect all defective produce

Automated grocery company could not accurately detect all defective produce without a large amount of labeled data. Defective data was especially difficult to collect.

Detection Before Advex (Left) and After Advex (Right)

SOLUTION: Synthetic Data For Defective Produce

With just 10 real images of a new part, Advex was able to create thousands of synthetic, labeled images.

Advex Generated Synthetic Data

RESULT: 100% of defects detected

With Advex, the company was able to detect all defective produce before they reached their customer.

Advex Case Study:
Faster New Part Support for Robotic Bin Picking

PROBLEM: Long Setup for New Customer Parts

Robotic solutions company could not accurately add new parts without significant data collection and fine-tuning.

Segmentation Before Advex (Left) and After Advex (Right)

SOLUTION: Synthetic Data For New Parts

With just 10 real images of a new part, Advex was able to create thousands of synthetic, labeled images.

Advex Generated Synthetic Data

RESULT: Customer supports new parts 5x faster

With Advex, the company was able to set-up new parts in 2 days instead of 11 days, saving time for their end-user and expensive on-site technician time for themselves.

Advex Case Study:
Higher Accuracy for Bin Picking In Warehouse

PROBLEM: Not All Objects Can Be Picked

A Robotic Bin Picking OEM could not pick all the different types of items that went through a warehouse, preventing them from closing deal.

With Advex, they were able to properly identify every part

SOLUTION: Synthetic Data for Infrequent Items

Advex automatically found gaps in customer’s dataset and generated 15k+ high-quality synthetic and labeled images

Advex Generated Synthetic Data

RESULT: 92% Accurate Bin Picking in 2 Days

With 92% accuracy picking parts, OEM was able to close deal

High variety leads to poor IoU accuracy, Advex remedies this with synthetic images (pictured left)

Proof of efficiency

Case studies

These stories showcase real achievements of our clients and demonstrate the effectiveness of our solutions in practical scenarios.

Truck Unloading

Higher Accuracy for Automated Box Unloading

Problem:

A Robotic Truck Loading/Unloading Startup needed better generalization to capture more customers.

Result:

With Advex, the company reduced their error rates by 25% in just 1 week which allowed them to promise higher starting accuracy and win more customers.

7 days

Time  
Investment

25%

Decrease in  
error rate

10x

Overall  
ROI

Download PDF

Truck Unloading

Faster New Site Calibration for Automated Box Unloading

Problem:

A Fortune 500 Automated Material Handling company developed an Automated Robotic Truck Unloading System but took 2 weeks to calibrate to every new customer site, limiting their ability to scale quickly.

Result:

With Advex, the company reduced their setup time form 2 weeks to 1 day which allowed them to scale 3x faster and make customers happier.

Download PDF