Data Annotation and Monitoring for A Smart Parking App Solution

Arranging and monitoring parking in big cities has been a prominent issue for decades, especially for countries with high automobile ownership rates. This unmanageability usually arises due to the lack of real-time information about vacant parking spots at any given time in any area. Vehicles that cruise around to locate a free parking spot lead to instances of additional traffic, higher emissions, and wastage of fuel, energy, and time.

SunTec India was delighted to be chosen by a client to assist their globally-applicable AI/ML-based Smart Parking Application to enhance its predictability and functionality.

The Client

This client is a Europe-based company with a penchant for creating simplified, automated, and smart solutions to everyday problems. They introduced and developed a smart parking system to monitor parking space availability in real-time.

The purpose of this smart parking application was to detect vacant parking slots, preferably in real-time, and provide end-users with access to that information while preserving its accuracy.

Their system delivers a suite of comprehensive solutions for easier parking, including overhead sensors monitoring, intelligent analysis of parking spaces, and real-time data reporting in the form of useful insights. The system used advanced sensors to identify free and occupied parking spaces across car parks, streets, motorway service areas, etc. The application is designed to inform the user of the nearest vacant parking lot with GPS directions. This results in increased ease, lower pollution, and happier visitors.

Business Problem

The company planned to mount sensors across different parking spaces, on walls, pillars, and street lamps, to get a complete and unobstructed view of all the parking sections and vehicles in the area. The sensors were designed to capture fresh images and relay them to the backend at a three-second delay. GPS mapping was associated with the sensors to coordinate the exact location of each free or occupied parking space in real-time to mobile and navigation devices, to facilitate hassle-free navigation. Additionally, the data collected from the sensor images, i.e., vehicle numbers, parking space usage records, payment information, etc. were to be shared with different city authorities.

The scope of this smart parking project required our team to achieve two major milestones.

1. Set up sensors at parking sites/streets, install them with Google’s geographical mapping, and test their output to ensure consistency.

2. Monitor live images streamed by the sensors, identify designated parking spaces, conduct image data annotation, and use the output to create a training data set for the AI-driven smart parking system.

Challenge Identification

At the time of project commencement, the smart parking was under development. There were many variables involved in the process, like coding slip-ups, bugs, redesigns, etc., which made up a motherload of potential problems. To combat this particular challenge, we put together a team of professionally adept programmers, AI/ML experts, and data annotators. We also scheduled training with the client and grasped the intricate aspects of the project.

We also faced several other challenges during the execution of this smart parking app project.

Proposed Solution

Our first approach was to amass a team of competent and tech-savvy resources and train them as per the client’s initial instructions. After identifying the most time-intensive challenges, we distributed the team into sections, dispatching each of them to take care of one part of the problem.

The client’s software solution was critically analyzed, the sensors were examined, and all the results were shared with the client. We provided a dedicated team of trained resources 24X7 to manage real-time data & images of sensors located at different time zones. This team monitored those images at a speed of 600 FPH. We also employed a backup team to eliminate any chances of delay in the processing of a new image on the system. We dedicated another team to take care of the sensor’s training data set and prepare it for better performance.

Final Result

Despite our initial inexperience of handling such a large-scale AI/ML project with multiple layers of requirements, we did our best and managed to salvage the situation and provide favorable outcomes to the client.

We helped the client achieve their primary goals-

All these results were accomplished while conserving the accuracy, consistency, and quality of data, without compromising on its time and cost-effectiveness. Our services were received with positive feedback from the client, who expanded the initial team of 5 team members in 2016 to that of 30+ team members dedicated to multiple tasks.

About SunTec India

At SunTec India, we understand the significance of data accuracy and quality on an AI algorithm’s predictions. Backed by a team of competent programmers, professional annotators, and data specialists, we have served numerous global organizations with tailor-made data analysis solutions, using our multi-dimensional operational approach to satisfy their use cases and ensure favorable outcomes.

To know more about how our data annotation services, reach out to us at info@suntecindia.com.