Image Labeling Service

Introduce service

FSI’s image labeling service is the annotation and description for images, which can then be used as AI training data for different machine learning fields such as self-driving cars, augmented reality. (AR) and face recognition.

FSI provides diverse image labeling service including:


  • 2D Bounding Box 

Our annotation specialists draw and label 2D bounding boxes around the objects specified in your image. Thereby improving the ability to recognize and improve the ability of machine learning to detect objects in autonomous vehicles (cars, drones), retail and e-commerce models when identifying clothes, accessories, furniture, computer vision models,…


  • 3D box / cubes

3D cube annotation asks the annotator to draw boxes around images to depict approximate length, width, and depth. Use our 3D cube annotation services to train object recognition on autonomous vehicles, augmented reality modeling, and other computer vision solutions.


  • Semantic segmentation

Detect and classify objects in a specific area within an image. Through a powerful technology platform used to annotate images in deep learning projects, FSI will meticulously annotate every object in the image with pixel-perfect precision, color-coding the image., and use its own semantic labels to help you solve problems such as: Detecting specific crops, weeds, or soil types in the field; …


  • Polygons

Annotate your images by drawing points around focus objects. This is useful for training object detection and localization algorithms, polygonal annotation is suited for annotating irregularly shaped objects, like street signs in traffic images or house in an aerial photographs.


  • Line & Splines

Identify and label lines and boundaries of multiple image types. With these services, we have promoted several vehicle recognition models with the ability to learn precise lane navigation by Annotating sidewalks and sidewalks, detecting different lanes on the road to simplify navigation for self-driving cars.


  • Text Annotate

FSI experts will tag text from various previously categorized images, and add the appropriate description with annotated text to facilitate fast data and more exactly. This service helps train the computer vision model to correct natural language interpretation, serve customer service automation, or teach ML / AI models to do meaningful conversations with people.


  • Video Annotate

We train your data by sorting, annotating objects moving in each video frame that have different lengths. The resulting data scan help AI / ML models train: Autonomous vehicles track moving objects, Track human activities and interpret postures and facial expressions, the Estimated motion of autonomous vehicles, drones, …


  • Point & Landmark Annotate

By plotting sequential dots at specific locations in an image, we create accurate data sets that can teach deep learning models about the trajectory of moving objects in an image and estimating the same with increased precision. Applications of this type of labeling include Face recognition, Position detection in the athlete for advanced sports analysis, Facial recognition, features and emotions, Precise gesture detection…

Our image labeling service will help you

  • Reduce implementation costs: Save the cost of setting up the organizational apparatus, the cost of management personnel, deployment personnel and machinery, supporting image labeling technology.
  • Save time on data entry: Reduce excess time in management processes, help businesses focus on key points in their business processes.
  • Improve business performance: Help businesses focus human resources and costs on key activities that bring profits to the company, instead of consuming manpower for labeling
  •  Multi-layer data security: FSI image labeling process meets ISO / IEC 27001: 2013 information security standard and ISO 9001: 2015 quality management standard

Advantages of the service

  • FSI provides professional image annotation services in multiple languages: Vietnamese, English, Japanese, Korean, …
  • Customizable image labeling process according to customer needs: 2D, 3D box labeling; label text; video tagging;…
  • Handling a large amount of work in a short time thanks to a large staff working in 3 shifts, continuously 24/7, thereby shortening the deployment time for customers.
  • Pictures are accurately labeled, ensuring the most stringent requirements of customers.