Pre-trained
deep learning vision models
Data is a critical step to building a vision model, but sometimes it is useful to skip the development and compute time needed for training. SBX can train custom models for a variety of vision tasks.
.jpg)
Why Vision Models Are Essential
Computer vision models are designed to translate visual data based on features and contextual information identified during training. This enables computer vision models to interpret images and video and apply those interpretations to predictive, decision-making tasks to quicken your workflow.
Instance Segmentation
The goal of these vision models is to detect visible sections of objects in an image.
2D Keypoint Detection
The goal of these vision models is to predict 2D location of object keypoints in image.
6D Pose Estimation
The goal is to predict relative position and orientation of each object. Deep learning models predicts 2D keypoints and 6D pose is recovered by 2D / 3D solver.
Ready To Use Computer Vision Models?
Witness the Possibilities of Synthetic Data Through Our Case Studies
Curious to see what synthetic data can do for you? From object detection to item manipulation and identification, view our case studies today to turn your business problem into a data-driven solution.
Vision Models FAQ
Should I use an in-house team for synthetic data generation?
SBX's mission is to deliver the value of synthetic data as quickly and affordably as possible. Creating high quality synthetic data requires a specific set of technology and skills. SBX was formed around this goal and our team of engineers and technical artists. Think of SBX as your CGI partner.
Should I start with real data?
If you are hesitant to begin with synthetic data it makes sense to see how far a small training set of real data can go. SBX can annotate small batch real data (see hybrid training page)
How much will it cost? How long will it take?
SBX typically aims to create custom synthetic datasets in 10-15 business days. Pricing includes all assets and benchmarking, and so can depend on the type of data being requested. Contact us for more info.
Will it work for my industry / problem?
SBX has built computer vision training sets for a variety of domains. Check out industries page to learn more.
Instance Segmentation
The goal of these models is to detect visible sections of objects in an image.
2D Keypoint Detection
The goal of these models is to predict 2D location of object keypoints in an image.
6D Pose Estimation
The goal of these models is to predict the relative position and orientation of each object. Deep learning models predict 2D keypoints and 6D pose is recovered by solving for correspondence between 2D + 3D keypoints.
Ready to use computer vision models?
Share 25 images from your vision system, and we will generate an optimized training set of 25,000 annotated synthetic images.