An Intelligent Surveillance System For CPEC Route

Project Title

An Intelligent Surveillance System For CPEC Route

Project Area of Specialization Artificial IntelligenceProject Summary

China-Pakistan Economic Corridor (CPEC) is a billion-dollar strategic project which is going to reshape the economic structure of Pakistan. Basically, CPEC is a sub-part Belt-and-Road Initiative (BRI), conceived by Chinese President Xi Jinping, which will link over 132 countries around the world. The core purpose of this initiate is to connect the Asian, European, and African countries through silk road in order to establish strong economic integration across the globe. In the near future, CPEC will be one of the central hubs of this initiative which will link European and middle east and Chinese industries through Arabian sea. Governments of both countries, the Pakistan and China are fully focused to accomplish the CPEC by 2030 to ensure the intercontinental trade through this route.

With the successful establishment of CPEC, Gwadar port (Baluchistan) will be linked with Khunjarab-Pass (Gilgit-Baltistan) via silk road which will pass through most areas of Punjab, Baluchistan, Khyber Pakhtunkhwa, Azad Kashmir, and Gilgit-Baltistan. Consequently, there will be tremendous increase in national/international industrial transport traffic on this trade route, which will ultimately require an appropriate surveillance system to ensure the security of traffic on this trade route. No doubt, there will be numerous vulnerabilities regarding the security of transportation because road will pass through crowded cities, agricultural villages, deserts, and hilly areas. These open questions which are an indication towards the need of an intelligent surveillance system to ensure the safe and secure transportation throughout the CPEC route.

To overcome the above-mentioned issues, we are aiming to propose a robust vehicle surveillance system to ensure the safe traffic movement and tracking throughout the CPEC route. The proposed will based on end-to-end re-identification method which will be capable of multi-dimensional vehicle detection and re-identification throughout the route. It is important to mention here that, the proposed system will act as central control system which will get visual stream from the active cctv cameras installed on the CPEC route. Resultantly, it will helpful in automated vehicle detection and tracking in terms arrival time at junction points and passage information.

Project Objectives

We are aiming,

1. To publish Asia's Largest Vehicle Re-identification dataset.

2. To develop a state-of-the-art surveillance system to ensure the vehicle detection and tracking throughout the CPEC route.

3. To develop an intelligent system to detect anomalous condition and activate befitting security alerts.

4. To establish "Made in Pakistan" intelligent surveillance system which can play an important role in increasing the economic growth of our country.

Project Implementation Method

The proposed system is an embedded system consists of both, the hardware modules which include imaging devices, processing units etc. and the software modules such as vehicle detection and tracking system, as shown in figure 1.

The step-by-step development procedure of proposed system is elaborated below.

1. Development of Vehicle Detection and Re-identification System

In the past couple of years, numerous deep learning feature-based vehicle re-identification system have been proposed. With the rapid advancement in computer vision and pattern recognition, convolutional neural networks-based vehicle re-identification methods have demonstrated better applicability in terms of real-time intelligent transportation systems. However, CNNs require piles of training data to maintain their accuracy in real-time applications. In this regard, many public vehicle re-identification datasets i.e., VRAI, and VRIC have been proposed to reshape the traditional intelligent transportation systems. But the matter of fact is that these datasets are constructed from cctv videos, which only provide top view of vehicle. Consequently, the other dimensional views of vehicles are ignored. Moreover, which do not cover the diverse road objects. Moreover, there classes are quite different from the road objects (i.e., vehicles, road infra-structure, and surroundings) of Pakistan. Consequently, the performance of the object detection system trained over the above-mentioned datasets will be greatly influenced while evaluating in Pakistani road environment. To overcome these issues, we are aiming to collect diverse driving video-based data form different provinces of Pakistan i.e., AJK, Punjab and KPK. Based on the collected videos, we will construct Asian's largest vehicle re-identification dataset, considering sub-continental vehicle classes in particular.

After completion of dataset construction, we will start data pre-processing in order to prepare dataset for annotation. Later on, we will perform frame extraction on the videos to get the useful images for annotation. After completing the data annotation, we will step ahead towards the development of convolutional neural network-based vehicle detection and re-identification. For this, we are aiming to present customized vehicle detector along with CNN based feature extractor to ensure the robustness of system. To avoid over-fitting, initially, we will train the purposed CNN on ImageNet dataset. Later on, it will be fine-tuned on our self-constructed dataset in order to achieve maximum accuracy. Simultaneously, we will implement Yolo-v4 to perform vehicle detection in order to achieve the desired output.

Benefits of the Project

The proposed system will help in,

1. Ensuring the 24/7 intelligent vehicle surveillance on CPEC.

2. Ensuring the automated vehicle flow monitoring and information collection i.e., Vehicle Paasge Time, Arrival time etc.

3. Ensuring safe and secure transit on CPEC route.

4. Portraying a positive image of Pakistan in international market in terms of security, which will ultimately increase the trustworthiness of Pakistani market in international market/investors around the world.

5. Improvement in economic growth of Pakistan.

Technical Details of Final Deliverable

The final deliverable of the proposed system will be an end-to-end intelligent vehicle surveillance system for CPEC route. The proposed system will be a generalized framework, which will be compatible with every operating system. The system will only require a processing unit with connecting imaging devices to ensure execution. It is important to mention here that, as the proposed vehicle detection and re-identification system will be trained on over 200,000 vehicle images (including multi-aspectual of vehicles) to ensure its robustness in real-time evaluation. The proposed system will be a 12v DC powered system, which will also help in saving electricity of Pakistan.

Final Deliverable of the Project HW/SW integrated systemCore Industry TransportationOther IndustriesCore Technology Artificial Intelligence(AI)Other Technologies Big DataSustainable Development Goals Industry, Innovation and InfrastructureRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	80000
Imaging Device (1080p, 4 MP, 30 FPS)	Equipment	2	15000	30000
Nvidia GPU enabled Processing Machine	Equipment	1	28000	28000
USB Extension Cables	Equipment	5	2000	10000
USB Adaptor	Equipment	2	1000	2000
Vehicle Overhead	Miscellaneous	4	2500	10000