4 CoreUse Cases
ContactGet Started

4 Core

AIMMO has End to End DATA Solutions

AI DATA Solution 4 Core is implemented in all industries that involve both people and AI, such as autonomous driving, Smart City, CCTV, Robotics, Drone, ships, and construction equipment. Improve the data-centric AI model’s performance with our 4-step AI model data solution of data collection, filtering, labeling, and evaluation.

core-intro-image
Core 1

Data Collection

AIMMO owns data collection vehicles for autonomous driving, which enables us to specify and simulate necessary situations for data collection. Our ultra-light Edge Device can trigger events like Scenario and ODD schedule, which prevents the collection of duplicate data. Ultimately creating an effective system that reduces the cost and time of our clients’ data collection.

Data Collecting Edge DeviceIoT Wireless N/W + AI chip
Core 2

Data Curation

AIMMO’s Data Curation reduces the data processing time by quickly filtering out customized data for each client. Not only that we are fast with our data extraction with ODDs or Scenarios, our extracted data can also be greatly customized for each client’s need. Therefore providing a specific, versatile data to use.

Filtered Data
Small DataODD(Operational Design Domain) : Classifying data with specific environmental ODDs, such as rain or snow.Scenario : Classifying data under a certain pre-set scenario.
Core 3

Data Annotation

AIMMO has Pre-trained Model (Smart Labeling) and Custom Model technologies. Our Pre-trained model can automatically label certain entities like people and cars, while our Custom Model can label even more detailed parts through a specialized learning model customized to your needs. For example, while the Pre-trained model can label people, a Custom Model can label specific parts of people, like hands or heads. At this point, we quality check with AI SQIP, which is faster than manual labeling and also serves as another round of quality check.

Smart Labeling
Clean DataPre-trained Model : Labels certain entities, such as people or cars.Custom Model : Can label specific parts of entities, customized to your need.
Core 4

Data Evaluation

Data Evaluation is one of AIMMO’s key techniques that runs multiple rounds of quality checks and evaluations used for consultations and feedback. In the process, our system catches gaps in data such as uneven distribution, unexpected situations, or irregular data. Then we use AI SQIP to evaluate whether to re-collect data or use synthetic data. Using synthetic data in this situation could be extremely effective since it builds the virtual environment and run the specific situations in need. It ultimately extracts the exact labeling inference needed. By going over all “irregular” situations, it provides accurate, high-quality data.

Robust Data Triggering(Edge case, Scenario, Data Balancing, Unusual Object, Data Distribution)Synthetic Data : An effective labeling inference obtained by building and running customized virtual situations.
before image
after image