Surveilia
Due to cheap technology, Surveillance cameras are installed everywhere such as streets, Shopping malls, Hospitals, Schools, Banks and even at residential area to observe human activities. The basic purpose of keeping track of human activity is to handle sudden events that may require the attention o
2025-06-28 16:36:12 - Adil Khan
Surveilia
Project Area of Specialization Artificial IntelligenceProject SummaryDue to cheap technology, Surveillance cameras are installed everywhere such as streets, Shopping malls, Hospitals, Schools, Banks and even at residential area to observe human activities. The basic purpose of keeping track of human activity is to handle sudden events that may require the attention of the public or maybe only of authorities. The traditional watch system is human-based so definitely full of errors, biases and more importantly exhaustive in nature. In the past, there are a lot of anomalous events which could be avoided but traditional surveillance system failed to do the job such as Puyallup South Hill Mall burglary and Westfield Century City Mall robbery and many others of similar nature. Pakistan is an underdeveloped country and needs attention in several departments and surveillance system is one of them.
The proposed solution is a complete product that is not only a product but a complete surveillance system that aims at providing maximum accuracy with minimum false alarms. The process of building the product involves incorporating the deep learning algorithms for activity recognition by taking advantage of a huge amount of data produced by CCTV cameras to provide a fully trained machine. The main purpose of the product is to assist humans in handling surveillance-based issues not to replace them completely. This product will be trained by using activity recognition dataset “something-something v1”, “something-something v2” and “UCF Crime” to perform deep video analysis, which detects the anomalies such as burglary, theft, vandalism, etc., in live streaming CCTV and alarms the security instantly. The most important aspect of the proposed product is its usability in every kind of sector i.e. Indoor/Outdoor and Public/Private.
Project ObjectivesThe main objectives are as follows:
- To build a product that can detect humans activities and their interpretation through automated video analysis using deep learning knowledge.
- The product should assist the security personnel so that the process of handling anomalies can be improved in real-time.
- To perform surveillance tasks as humans do
- Reducing the “False Alarm” rate to the minimum to increase the reliability of the product
The product will be build by using the Agile Approach so that the new requirements and issues can be handled in a very efficient manner to achieve real-time surveillance. Moreover, Python, which is a very famous scripting language will be used to incorporate deep learning algorithms to achieve efficiency in the job. We will also be using Python for development of front-end of application and the back-end of application.
The second important aspect is the dataset for the problem and for that video-based activity recognition dataset will be used to train the model for the ultimate product. The famous deep learning algorithm such as C3D will be used to extract features from the video and then will be fed to the model for the training phase. The training model will be tested to measure the performance and then will be deployed on jetson Nano and Raspberry Pi which will be used as a product for the surveillance system and will be operated on a Live camera stream to detect anomalies to generate alarm, accordingly.
Benefits of the ProjectThis project is aimed at providing the user-friendly application which will be helpful in managing surveillance tasks by automatically detecting anomalies and hence our all security system will be improved. The product will benefit through
- Reducing Human efforts as it would save the exhaustive task of 24/7 watch
- Real-time spotting anomalies and other related events
- Identifying the suspects and real victims
- Reducing crime rates by introducing strict checks everywhere
- Live Monitoring of targeted area/Location
- Deploying the product nationwide
The final deliverable of the project is an application that would be ready to use the form and can be used in every kind of sector where surveillance is required. The product is aimed at providing full-time monitoring of CCTV cameras by eliminating the need for continuous watch and will produce alarms for anomalous events so that the respective person can address the issue.
Final Deliverable of the Project Software SystemCore Industry SecurityOther IndustriesCore Technology Artificial Intelligence(AI)Other TechnologiesSustainable Development Goals Sustainable Cities and Communities, Peace and Justice Strong InstitutionsRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 74513 | |||
| Raspberry PI 4 Model with 4GB RAM | Equipment | 1 | 14500 | 14500 |
| Papago GoSafe 220 Dash Camera | Equipment | 1 | 24999 | 24999 |
| Nvidia Jetson Nano Developer Kit | Equipment | 1 | 28514 | 28514 |
| Printing | Miscellaneous | 1 | 3500 | 3500 |
| Stationery | Miscellaneous | 3 | 1000 | 3000 |