Train custom Neural Networks

Why train models in Supervisely

• Get started in one click: don't waste dozens of hours figuring out how to install, configure and adapt another model from the Internet - we've done the hard part for you.

• Convenient user interface. With the simplified way we create interfaces using Supervisely Apps, we can add a GUI to make it easier to interact with the integrated code.

• Deployment on any computer. Run the auto-generated command in a terminal to launch Supervisely Agent and simply deploy new training tasks with a few clicks.

• We work together as a team. Because these models and tools now work online instead of on your personal computer, you can easily share and collaborate with your team members.

• Integration with other applications: a project converted into an application has access to the entire ecosystem: import from different formats, visualization, exploration and much more!

Some examples of the train dashboards:

• Train Detectron2

• Train MMSegmentation

• Train YOLOv8

• Train MMClassification

• Train UNet

---

Step-by-step tutorials on training models in Supervisely:

• No-code tutorial: train and predict YOLOv8 on custom data

• How to Train Smart Tool for Precise Cracks Segmentation in Industrial Inspection

• How to Train a Model with Only 62 Labeled Images using Semi-Supervised Learning