Data Annotator – Image, Video, LiDAR & DICOM Annotation

Company: Obraz Private Limited
Location: Mannarkkad, Palakkad, Kerala
Employment Type: Full-Time
Experience: Minimum 2 years of hands-on data annotation experience (2D image, video, LiDAR/3D point cloud, or DICOM medical imaging), including practical use of industry annotation tools

About Obraz

Obraz Private Limited specializes in developing secure, production-grade AI and Computer Vision solutions for enterprise and defense applications. Our work spans areas such as defense and surveillance, healthcare imaging, industrial automation, and smart infrastructure.

About the Role

Obraz is looking for detail-oriented Data Annotators across all core modalities — 2D image, video, LiDAR/3D point cloud, and DICOM medical imaging — to join our annotation team and support AI/ML and Computer Vision projects.

In this role, you will work within your specialization to identify, classify, and label objects and regions accurately. Depending on your modality, you will create 2D bounding boxes, polygons, keypoints, and semantic or instance segmentation; frame-by-frame video object tracking; 3D bounding boxes and point cloud segmentation; or region-of-interest annotations on DICOM medical imagery, and ensure annotation quality in accordance with project-specific guidelines.

This is an opportunity to contribute to AI development while working with multiple data modalities and production-grade annotation workflows.

Tools & Platforms

You will work on our proprietary in-house annotation suite (2D Annotation, Video Studio, and 3D Point Cloud workspaces, with AI-assisted pre-labeling and built-in QA modules); full training is provided. Prior hands-on experience with any of the following industry tools is strongly valued:

• 2D Image Annotation: CVAT, Labelbox, V7 Darwin, Supervisely, Encord, Label Studio, Segments.ai, Roboflow.
• Video Annotation: CVAT (video mode with interpolation), Encord, Labelbox Video, V7 Darwin, Supervisely Video.
• LiDAR / 3D Point Cloud: Segments.ai 3D, BasicAI Cloud, Supervisely Point Clouds, Labelbox 3D, CVAT 3D; point cloud viewing/QC in CloudCompare; familiarity with PCD, LAS/LAZ, and PLY formats and camera–LiDAR projection workflows.
• DICOM / Medical Imaging: 3D Slicer, ITK-SNAP, MITK, OsiriX / Horos, RadiAnt DICOM Viewer, MD.ai, Encord DICOM; familiarity with DICOM standards, series navigation, and windowing.
• Output Formats: COCO, YOLO, Pascal VOC, PNG/mask exports, and point cloud / track metadata in JSON or CSV.

Key Responsibilities

• Annotate data within your specialization (2D images, video sequences, 3D LiDAR point clouds, or DICOM medical imagery) using project-specific tools.
• Create and adjust 2D bounding boxes, polygons, polylines, keypoints, and semantic/instance segmentation for image and video projects.
• Create and adjust 3D bounding boxes (cuboids) for vehicles, pedestrians, cyclists, and other relevant objects, and perform point cloud segmentation and tracking where required.
• Annotate regions of interest, anatomical structures, and findings in DICOM medical imagery in line with clinical and project guidelines.
• Perform frame-by-frame object tracking and maintain consistent object identities and labels across consecutive video frames.
• Follow guidelines for object dimensions, orientation, visibility, occlusion, and class hierarchies.
• Review annotations and correct missing labels, classification errors, and inaccurate boundaries.
• Operate annotation platforms efficiently, including tool-side settings, hotkeys, interpolation/tracking assists, and export configurations.
• Incorporate feedback from quality reviewers and team leads.
• Meet daily productivity targets while maintaining annotation accuracy and quality.
• Communicate unclear cases and project-related queries promptly.
• Follow company requirements for data confidentiality and information security.

Required Qualifications and Skills

• Minimum 2 years of hands-on annotation experience in at least one of the following: 2D image, video, LiDAR/3D point cloud, or DICOM medical imaging.
• Practical working experience with at least one industry annotation platform (e.g., CVAT, Labelbox, V7 Darwin, Supervisely, Encord, Label Studio, Segments.ai, BasicAI) or medical imaging tool (e.g., 3D Slicer, ITK-SNAP, OsiriX/Horos, RadiAnt, MD.ai).
• Understanding of annotation techniques relevant to your specialization (bounding boxes, polygons, segmentation, keypoints, 3D cuboids, object tracking).
• Strong attention to detail and visual interpretation skills.
• Good computer skills and familiarity with annotation tools and workflows.
• Ability to follow detailed project instructions consistently.
• Good communication, teamwork, and time management skills.
• Willingness to learn new tools and incorporate feedback.
• Commitment to maintaining quality standards and meeting deadlines.

Good to Have

Experience in any of the following will be an advantage:

• Multi-tool and multi-modality experience (e.g., CVAT plus Labelbox, or image plus video, or camera and LiDAR sensor fusion).
• Autonomous driving datasets (vehicles, pedestrians, lanes, ADAS validation) annotated in 3D point cloud tools.
• Medical or clinical imaging annotation (radiology, pathology) using DICOM viewers and segmentation tools.
• 3D object tracking, point cloud segmentation, and sensor fusion projection workflows.
• Annotation quality checking, QA, or review using tool-side QA modules or consensus/IAA workflows.
• Experience with model-assisted pre-labeling (e.g., SAM-based masks, auto-tracking) and correcting machine-generated labels.
• Working with large-scale datasets and production annotation pipelines, including standard export formats (COCO, YOLO, VOC).

What We Offer

• Opportunity to contribute to AI/ML and Computer Vision projects across multiple industries.
• Exposure to advanced annotation workflows across image, video, LiDAR, and DICOM modalities.
• Training and certification on our proprietary annotation platform, including AI-assisted pre-labeling and QA workflows.
• Collaboration with experienced annotation teams and project leads.
• Opportunities to strengthen technical and quality review skills, with growth paths into QA and team lead roles.

Who Will Succeed in This Role?

The ideal candidate is observant, patient, consistent, comfortable working with complex data modalities (2D, video, 3D, or medical imagery) and the tools that support them, and committed to delivering accurate, production-ready annotations at scale while adapting quickly to evolving project guidelines, ontologies, and platforms.