It’s common at the end of the year to step back and take a look at what you’ve accomplished. 2021 has been a very proud year in the advances of our products with our slew of improvements, enhancements, and additions. In case you missed out on anything, here is a reminder of what Dataloop unwrapped in 2021.
Dataloop New UI Facelifts
If we had to sum up 2021 in one word when it comes to Dataloop, it would be facelift. This past year we updated and improved our analytics, video, and image platform.
Analytics Dashboard
Every team that is managing unstructured data knows that this is a complex process with many factors. How do you track annotator performance or ensure the labeling instructions you provide are clear? How do you analyze the time to annotate and in turn identify areas for improvement? Is your workforce up to par with your industry standards?
Dataloop set out to answer all these questions with the new and improved Analytics Dashboard giving you a birds-eye view of how your team is performing, allowing you to find anomalies at a glance.
Our Analytics Dashboard makes it easy for you to see the progress with a label status showing you the number of annotations from each label, and performance dispersion. While status information is available in task-context only, the label information is available on a dataset and project level. Each action is measured and analyzed, helping you reach data-driven optimizations at every step of the process. Gain full transparency in your annotation workflows in real-time in a seamless and faster output time. This allows you to have easier collaboration with your team, with more automation essentially optimizing your productivity.
Perhaps most importantly, the Analytics Dashboard gives you full insights into your teams’ performance. You can compare annotation quality, time to label, any issues identified, and much, much more.
Video Platform 2.0
Video Platform 2.0 is designed to offer faster and more accurate video annotations. Gain better control of lengthy video files with the ability to move and resize the window, drag and drop the marker to change your position on the timeline. Navigate your videos more efficiently by jumping from frame to frame easily (this feature is accessible from multiple locations). At any time click to where you’d like to go on the timeline or drag and drop the current-time indicator. Select a keyframe range from the list and jump to the keyframe’s first frame. You can also use the skip and rewind button or keyboard shortcuts for a much quicker and more seamless process. During QA tasks, you can now experience a faster and easier process. Click on any keyframe from the list to jump to where the change occurred. This allows you to quickly go back to the frame you were working on previously. Additionally, you can review the annotations mapped in order to validate the existence of video labels giving you better orientation and options to jump directly into scenes without annotations.
Image Annotation Platform 2.0
Dataloop’s Image Annotation Platform 2.0 enables faster and more accurate annotation work than ever before. With a more intuitive and fresh look, it’s designed to accelerate your annotation work. We’ve improved item load speed, zoom speed, and many other performance factors, for a faster and smoother experience. Effortlessly clone selected annotations with just the click of a button. Take advantage of the new ontology logic with attributes that can be added directly from within the platform. An improved annotations list was rebuilt to simplify your work, making it easy to access hundreds of annotations, groups, filters, and apply bulk actions. Group by label, free text search, auto annotations numbering, parenting hierarchy, and many more.
Introducing LiDAR Annotation
Dataloop now supports LiDAR annotations. Speed up your LiDAR annotations with sensor fusion, cuboid annotations, topographic depth view, and point cloud focus. Our high-performance tool is easy to set up that can effortlessly handle the abundance of incoming data, allowing you to annotate quickly. Our dedicated point-cloud annotations for multi-dimensional analysis of objects in a given space enable you to measure the distance, depth, position, and surroundings of each annotated object. Dataloop’s cuboid annotation tool can give you a full picture of depth and space in order to mass label multiple objects in 3D within the 3D point cloud and allow you to gain granular insights into your scenes.
Automation Pipelines
With Dataloop’s automation pipelines we’re aimed at saving you time, but we’re also allowing you to gain so much more with your data. We accomplish this by processing your data using machines and human annotation. We provide effortless automation for the movement and transformation of your data. Connect Dataloop’s components to enable a smooth, automated flow of data from one station to the next. Pipelines allow you to better manage your tasks and focus your efforts where they’re needed most.
We support built-in templates giving you the ability to build your own pipelines. You can create custom data automation pipelines using a no-code drag-and-drop interface or through a developer-friendly Python SDK to weave together human and machine workflows. Run fully integrated compute workloads and seamlessly integrate ML models using comprehensive developer tools and a comprehensive developer toolset. Develop your ML pipelines or create human-in-the-loop data validation to scale production. We’re excited to roll out the pipelines which are currently in beta version.
Automation Tools
With the utilization of automation tools and active learning, you can boost the manual annotation processes. With automation tools such as auto-segmentation, you can accelerate your semantic segmentation tasks. How would you like to gain 100% pixel enforcement with Dataloop? We bring semantic segmentation to a whole new level. This unique feature ensures that items are segmented completely, and no pixels are left unmasked. Annotators can use the “unmasked pixels” validation tool to highlight all of the unmasked pixels left to reach 100% of the item.
Here are some other great automation features we added this year:
4-Points to Polygon Magic Tool:
Automatically convert a simple 4-Points polygon into a complicated tight contour polygon around an object, by using the Polygon to Polygon 4-Points Magic Tool.
You can automatically generate a polygon around the object’s edges. Adjust and edit this polygon using the polygon edit mode.
Box to Semantic Tool:
Semantic is all about the speed of converting boxes to semantic masks while still preserving accuracy. In addition, we’ve added great tools to support the semantic labeling flow with an eraser on the pixel level, the ability to zoom 30x into the object for more accurate semantic labeling/erasing, and conversion tools like polygon to semantic, box to semantic, etc.
Preparing for 2022
As 2021 wraps up, everyone is already in the mode of preparing for the year ahead. Planning, highlighting their top priorities, assessing what worked this year, what didn’t, and overall taking inventory. If you’d like to learn more about these features and see how they could help your organization we’d love to talk, set up your 1:1 session with our experts today.
