Now here's an intelligent traffic analysis tool that can ease the task of town/traffic planners tremendously. Named TRAZER, the platform has the potential to make traffic flow smoother and faster, while making cities safer and less polluted.

Someone has rightly said, "There are no traffic jams when you go the extra mile." Considering the worsening traffic scenario on most Indian roads, it has indeed become essential to traverse that extra mile and look for solutions that could ease traffic bottlenecks.

In the last decade, traffic planners, city administrators and urban development consultants alike have been faced with the daunting task of finding ways to address the traffic menace. But with TRAZER, a traffic analyser and enumerator, they can heave a sigh of relief, as it is finally possible to systematically and comprehensively plan for and manage the traffic conditions for any city.

Developed by Noida-based KritiKal Solutions Pvt Ltd (http://www.kritikal.in), TRAZER is an offline video analytics-based product for analysing traffic conditions. Based on KritiKal's proprietary computer vision and image-processing algorithms, it is useful for traffic flow pattern analysis, intersection studies, and other traffic research areas.

TRAZER works on a video feed to detect vehicles, classify them into various categories, and track a whole lot of statistics, like instantaneous and average velocities, congestion in terms of queue lengths, throughput, etc, data that is currently computed manually. The information thus generated is also in a user-friendly format. TRAZER also provides a video backup to verify the findings at a later date.

A perfect solution for the Indian traffic scenario

Elaborating on the relevance of TRAZER in the Indian scenario, Jatin Sharma, CEO, KritiKal Solutions Pvt Ltd ( This e-mail address is being protected from spambots. You need JavaScript enabled to view it ) says, "The uniqueness of Indian traffic conditions poses a challenge to all those involved with traffic or town/city planning. On Indian roads, the traffic pattern is usually not uniform, and traffic density is quite high. Then there is low compliance with traffic rules and regulations.

"Besides, there is a wide variety of vehicles (including human traffic), a vast range of vehicular velocities (pedestrians, bicycles, LMVs/lightweight multipurpose vehicles, HMVs/heavy motor vehicles) and poor lane discipline (partially resulting from the first two factors).

"These issues make the traffic situation extremely complex, and make the task for any traffic analysis solution a whole lot tougher."

But Sharma asserts that TRAZER has been designed keeping the Indian (or developing countries, in general) kind of scenario in mind.{quotes}With TRAZER, city planners and traffic consultants can analyse offline traffic videos, study the statistics that are captured from the analysis, and use the data to plan traffic conditions for any city in a more systematic way,{/quotes}he adds.

Sharma declares that there are very few products worldwide that use similar approaches, but they also are meant for different applications. He reveals, "For example, there is a French company which uses similar technology for incident detection solutions on highways, in tunnels, etc. Another product, using a similar technology engine, focuses on preventing clashes at red light intersections. But these solutions are not applicable in India or any developing country, nor do they deliver the same functionality as TRAZER does."

In India, too, the tools and processes used currently for traffic analysis are limited. Says Sharma, "A manual observation process is used while profiling traffic, where people are deployed to monitor the traffic and generate statistics, etc. Also, there are some tools available based on two other types of sensors, viz, a magnetic loop sensor or piezoelectric pressure transducers. However, in both cases, sensors have to be embedded in the road for counting purposes, and their results are questionable in the kind of heterogeneous traffic conditions we have here."

TRAZER has been designed with inputs from TRIPP (The Transportation Research and Injury Prevention Programme), IIT Delhi, keeping the above-mentioned conditions under consideration and has been extensively tested.

What drives the TRAZER?

The basic technology behind TRAZER is object identification and classification in images/video, using learning techniques. The identification and classification happens through three simple steps:

- The system is initially trained on a set of inputs for the objects of interest.

- The system then applies this learning to a new input by itself, to efficiently identify the learned patterns and classify them appropriately.

- Once detected and classified, the object is tracked in subsequent video frames, to avoid multiple counting of the same object.

The actual input is in the form of a running video, so TRAZER does detection and classification on the image frames of the entire video. The output of this step is a trajectory of movement of the object.

(Refer to the snapshot of TRAZER software, given on page 33, which shows vehicle detection and classification results in a particular frame of a traffic scenario video.)