Training Tutorials
We have prepared several Jupyter notebook tutorials to illustrate how to train AI models in our ecosystem. You can find them stored in the depthai-ml-training repository on GitHub. The easiest way to try them out is by using the Google Collaboratory where all the notebooks are already loaded and prepared for running. Google Collaboratory allows you to use their fast GPU instances and, in some cases, even using the Google TPU (depending on the dependencies of the training frameworks). Alternatively, you can run the tutorials locally using your own resources.We divide our tutorials into two groups: the ones using our Luxonis Train library and those using different training libraries (also check Ultralytics integration docs). Our primary goal is to provide comprehensive tutorials covering the most common use cases for training neural networks for usage on DepthAI devices. Our tutorials cover the following tasks:Classification
Image classification is the task of assigning a predefined label or category to an entire input image.Luxonis Train TutorialsOther Tutorials
Object Detection
Object detection involves identifying and locating multiple objects within an image and providing bounding boxes around them.Luxonis Train TutorialsOther Tutorials
Semantic Segmentation
Semantic segmentation involves classifying pixels in an image into a predefined category without distinguishing between separate instances of the same class.Luxonis Train TutorialsOther Tutorials
Instance Segmentation
Instance segmentation extends semantic segmentation by distinguishing between individual instances of the same object class and providing a pixel-wise mask for each one.Luxonis Train TutorialsOther Tutorials
Instace Pose
Instance pose detection combines object detection and keypoint detection by localizing individual objects and predicting keypoints for each instance separately.Other Tutorials
