Video Annotation

TELUS International provides custom video data training datasets to help you develop your computer vision products. Use our video training datasets to improve the object detection and movement prediction features in your models.

Let our expert annotators draw and label 2D bounding boxes around multiple objects of interest in your videos. Improve object detection in autonomous vehicles, drones and more with our precise bounding box annotation within every frame of your video.

3D cuboid annotation requires our annotators to draw boxes around objects to depict length, width and approximate depth, as well as tracking of the objects across multiple frames of video.

Separate every pixel of an image into the segments you need your computer vision algorithm to recognize. Using our proprietary platform and tools, our annotators will help build video datasets with meticulously annotated objects and pixel perfect accuracy.

Using our high-precision labeling tools, we can visualize, label and track objects across frames in 3D point clouds for all types of lidars. Our proprietary AI platform covers auto-ground segmentation, accurate 2D-3D linking, auto-detection of static objects and ego-motion compensation.

We offer fast, efficient keypoint annotation over multiple frames for a range of use cases, including facial recognition, emotion detection and counting applications. Our high-precision landmark tools determine shape variations of minute and large objects via a sequence of points. We also provide pose-point annotation for human figure detection and movement prediction.

Ensure pixel perfect polygon annotations drawn around objects of interest in your frames. Useful for training object localization and detection algorithms, polygon annotation is well-suited for annotating objects with irregular shapes, like street signs in traffic images or houses in aerial imagery with a high degree of accuracy.

Using smart and intuitive line annotation tools, we accurately define lane lines in drivable areas for vehicle perception models. Our accurate lane detection services are helping to build the future of driving across cities, towns and uneven ground scenarios.

We support accurate video frames classification for autonomous vehicle perception models. Time stamps are tagged for attributes like time of day, weather, environment, object/border occlusion, and more.

Convert what is spoken on video into written text for subtitling or captioning. Video transcription makes your online videos more searchable and accessible because it provides a better UX and boosts SEO.

Video data contains a lot of useful information for companies today, from identifying car license plates to watching an active audience for signs of boredom. Whether you have a one-time project or need to analyze data on a regular basis, our experienced project managers will ensure a smooth engagement with precise results.