Video Examples

Why describe all the features when you can show them?

Video Labeling

Heart segmentation on Ultrasound Videos with Supervisely and Custom Smart Tool

Discover how AI assistance and tailored tools can tremendously speed up the labeling of medical data.

How to use Supervisely & Blender to train car parts segmentation Smart Tool

A lot of insurance companies who want to automate defect inspection and detect various problems need to find separate car parts first.

E-Commerce items segmentation on Videos with Supervisely and Custom Smart Tool

Background removal from ECommerce goods on videos allows to present those goods in a new, visual appealing fashion. It’s hard and time consuming to remove background from an image, so when it comes to a video this operation might be infeasible.

Plants segmentation on videos with Supervisely

Manual pixel-wise segmentation is time consuming for images but even harder and often infeasible for videos. Here we showcase how to segment plants and parts of plants on videos.

Cars tracking on videos with Supervisely

Let’s track cars on a video. Not one by one, but rather a bunch of them at the same time. So first the cars of interest are selected, then we track all of them. In the process, if necessary, adjustments are made.

Microorganisms tracking on microscopic videos with Supervisely

Let’s illustrate how microorganisms might be tracked on microscopic videos. In this demo, the task is addressed with a general purpose tracking model.

Image & 3D Labeling

Tags for image labeling. An overview

Video demonstrates the way Tags might be used in the labeling process. Image and object tags are explained as well as various types of tags, including text, numeric and “one-of” tag types.

3D labeling overview

Import of Lidar data (pcd files), Labeling with cuboids, Copying labels from previous point cloud to the next one, Using tags to assign attributes to objects.

🛠️Basic annotation overview

More labeled data you have — better AI you get.

Collaboration

Annotation at scale overview. Use case 1

Video demonstrates how to organize the annotation process at scale, using a toy example. Two users are assigned to complete simple annotation tasks. The first user is asked to label lemons while the second user labels kiwifruits.

Annotation at scale overview. Use case 2

Video demonstrates how to organize the annotation process at scale, using a toy example. Two users are assigned to complete simple annotation tasks. Annotation tasks are created by evenly distributing images from a selected project among users.

Exams to evaluate annotators performance

Video demonstrates how to create an exam for labelers and evaluate their performance.

Data Management and Stats

Data management part 1. An overview

The video provides an overview of data management tools available in Supervisely. Simple approach to data versioning and sharing is demonstrated via dashboard tools.

Data management part 2. Data Commander

The video demonstrates how to use Data Commander to address data management tasks once the data is imported to Supervisely. Essentially, Data Commander treats teams, workspaces, projects, datasets as directories and images / videos / point clouds with corresponding annotations as files.

Statistics and data exploration on images

Basic stats tools, Basic Data Exploration tools, “Classes stats for images” app, “Object Size Stats” app

Training Neural Networks

🔥Segmentation with custom neural network on public dataset

Automatic segmentation with custom neural network.

Import custom neural networks, train and apply to your data

Model Zoo is an awesome thing, but what if you want to use own model?

🤷👩‍🔧👨‍🔬Human Instances Segmentation (Faster RCNN + UNet)

Let’s consistently apply object detection and segmentation models to segment person instances.

Smart Tool

Smart tool overview

Put rectangle and we will pixelwise segment dominant object inside.

🌈Trainable smart pixel segmentation tool

It's just another neural network. And that means we can train it to work with any type of object, even the most complex one!

Human-in-the-loop

🤖Human in the loop for Person Detection

Let’s utilize Deep Learning to annotate thouthands of images in hours!

💊Medical image analysis with Supervisely

How to build blood vessel segmentation in retina images when you only have 6 images in training set.

Full-length Tutorials

🍋⚔️🥝Lemons vs Kiwi: solving semantic segmentation task

Entire process of building a prototype: semantic segmentation model is built to distinguish lemons from kiwi.

Python SDK

Python SDK, quick start with Jupyter Notebooks

The goal of the video is to illustrate how to start working with Python SDK in the shortest time period possible. To achieve that built in Jupyter Notebooks are used to demonstrate import and data management automation.

Trusted by Fortune 500. Used by 25,000 companies and researchers worldwide

Let's talk about how Supervisely can help you develop high-quality annotations now.