AI Verse - Resources
Our products create unbiased, labeled, synthetic datasets ideal for training top-performing Computer Vision AI models.
Discover how synthetic data revolutionized our tank detection model training.
Training a tank detection model using conventional data presents several challenges. One of the biggest obstacles is the scarcity of labeled data. Tanks are not everyday objects, and acquiring enough annotated images for training is extremely difficult due to confidentiality of images.
Synthetic
Datasets
Real World
Images
Speed, Cost, Flexibility
You can build a synthetic dataset for a fraction of the cost of a real-world image dataset. A 3D scene and a fully labeled image matching your use case are produced in seconds. Easily extend your dataset to match each new edge case throughout your development cycle.
Data Collection
Even if possible, in most cases, collecting real-world images is a daunting task. Privacy issues may also complicate the process. Procedural generation of synthetic datasets is a game changer. You create your own images in a few clicks and avoid any privacy issues.
Labelling
You can build a synthetic dataset for a fraction of the cost of a real-world image dataset. A 3D scene and a fully labeled image matching your use case are produced in seconds. Easily extend your dataset to match each new edge case throughout your development cycle.
Optimization
Even if possible, in most cases, collecting real-world images is a daunting task. Privacy issues may also complicate the process. Procedural generation of synthetic datasets is a game changer. You create your own images in a few clicks and avoid any privacy issues.
Winner:
Synthetic Datasets!
The Benchmarks prove it
Research Summary
To evaluate the efficiency of synthetic datasets to train a model, we conducted a series of benchmarks, comparing trainings done with synthetic images against trainings done with real-world images (COCO dataset). As of today, the results were established for two different models (Yolo V5 and Mask R CNN), for three different tasks of increasing difficulty (sofa, bed and potted plant detection). We conducted these tests with a 1000 assets in our database.
Procedure
Real-world image training datasets were extracted from MS Coco (HERE) for each class of interest. We obtained 3682 images containing the label “bed”, 4618 containing the label “couch” and 4624 images containing the label “potted plant” from MS Coco.
For each test, we used our procedural engine to generate a synthetic dataset. For “beds” detection, we used a 63k synthetic dataset, for “couches”, 72k synthetic images and for “potted plants”, 99k images.
We also used Imagenet (HERE) for pre-training models in several experiments.
Validation Datasets were constructed for each class of interest from OpenImage (HERE). We extracted 199 images containing the label “bed”, 799 images for the label “couch” and 1533 images for the label “plant”.
Conclusions
The domain gap between training sets and validation sets or live images is not exclusive to synthetic datasets. It is a general issue which also exists from real images to real images.
In fact, synthetic images are generally more efficient than real images for training models. This might seem counter intuitive because synthetic images are less realistic than real images.
However, image realism is not key to train a model due to the domain gap. Variance and distribution of the parameters are the crucial factors to obtain a model which generalizes well.
Variance and distribution of parameters are not easily controllable with real images.
Models may be successfully pre-trained on synthetic images and fine-tuned on real images or the other way round. It depends on the task and on the model.
BEDS
AI Verse Synthetic Dataset Sample Images
Bed: RCNN
Bed: YOLO
PLANTS & COUCHES
AI Verse Synthetic Dataset Sample Images
Potted Plants: RCNN
Potted Plants: YOLO
Couch: RCNN
Couch: YOLO
