Expanding Vehicle Detection Model

The Challenge

Big cities are synonymous with traffic jams that often make time wasted. Countless solutions have been deployed by the Government. One of them is how to make a system that can detect the density of vehicles on the road through public CCTV. This solution enables the Government to seek traffic spot so that  officers can be deployed to help unravel the congestion, beside that the flux of data  could also  serve as a data-driven insights for future policy action.

In order to work at its best, an AI solution requires a lot of data. Making training data becomes a challenge because of the many CCTV points that need to be validated and the density of objects in the image.

The Solution

Step 1: Image Cleansing

We received images from over 2000 CCTV points. These images have many variations in terms of time taken and the density of objects in the picture.

At first, the raw data is still mixed among those  that needs to be annotated or not by virtue of vehicle images on the captured data. For this, we utilze AI assistance to validate whether there are objects in the image that need to be annotated or not. This results in a more effective data selection exercise for subsequent annotation process.

Step 2: Image Annotation

Once the data are gathered, our automated annotation platform classifies images based on the time-taken to soothe the process so that our workforce can perform annotations accordingly.

To accelerate the annotation process, our platform generates predefined bounding boxes in advance, so the annotator could exercise their duty in a much faster manner as they  only needs to validate the results of annotations that already exist.

Each labelled data will then perform multistage human-based quality control. The aim is to ensure that each annotated image matches the intended quality

The Result

Acquaire can accelerate the CCTV image selection process to be 2-3 x faster than the traditional process. Our multistage quality control ensures the density of objects in the image does not reduce the quality of the overall annotation results. 

The annotated data that we provide allows our clients to obtain high quality training data in order to create AI solutions, hence help them in delivering real impact on the problem faced,  which is very crucial in use case such as smart city solutions and others.

Let's discuss about your project

About Us

AcquAIre is a data annotation service from Nodeflux which focuses on producing high quality data for the development of Vision AI technology.

Here for you

(021) 22718184 or hi@nodeflux.io

Where to find us

Jl. Kemang Timur No.24, RT.7/RW.4, Bangka, Mampang Prpt., Jakarta Selatan, 12730