Human Activity Recognition using Deep Learning
Most computer vision applications such as human computer interaction, virtual reality, security, video surveillance and home monitoring are highly correlated to Human Activity Recognition (HAR) tasks. This establishes new trend and milestone in the development cycle of HAR systems. The problem is th
2025-06-28 16:27:43 - Adil Khan
Human Activity Recognition using Deep Learning
Project Area of Specialization Artificial IntelligenceProject SummaryMost computer vision applications such as human computer interaction, virtual reality, security, video surveillance and home monitoring are highly correlated to Human Activity Recognition (HAR) tasks. This establishes new trend and milestone in the development cycle of HAR systems. The problem is that most applications based on HAR available are low in accuracy due to which the percentage of faulty recognition of human activities and false alerts is quite high.
Human activity recognition systems attempt to automatically identify and analyse human activities using acquired information from various types of sensors. Therefore, we are proposing an idea for a mobile application which helps recognize human activities. Our mobile application will alert the user if there’s any suspicious activity. User will also be able to monitor and analyse other person’s activities through video surveillance. Application will be able to recognize and classify several group of activities. This application will be developed using flutter and ML/AI kit.
Project Objectives• To develop an app that will recognize human activities in real time functionalities.
• To make an app that will alert users for suspicious activities.
• To provide users better surveillance.
Project Implementation MethodWe have decided to follow agile scrum methodology. With the help of agile scrum methodology, we will be able to achieve high accuracy and high-quality product. In the Specificity of the order of the technique we have around 7 steps which are mentioned below.
(i) Data Acquisition (Data Set for Human Activity Recognition): In this stage, we will collect data from relevant sources, pre-process that data and use the data for further mechanisms.
(ii) Construct a Model: The next stage involves construction of our model. This stage is crucial because our whole application depends upon how we train the model and what resources do we use.
(iii) Investigate transfer learning models: Transfer learning is when elements of a pre-trained model are reused in another model. In this stage, we will figure out what transfer learning models are best for our application.
(iv) Testing and training process: Testing stage is when we test and train our model’s accuracy in order to move on to development stage.
(v) Statistical analysis and state of art comparison: In this stage, we will compare all the trained models on various datasets and choose the one with the optimal parameters.
(vi) Design and development of application interfaces: In this stage, we will design user interface of our application and develop our crossplatform application in flutter.
(vii) Deployment of trained model on smart app: At last, we will deploy our application after proper training and testing.
Benefits of the ProjectSecurity is a mainstream need of human life in current era. The proposed project has multiple applications in this domain. It could be used for survillance purpose at factory premises, homes, offices, university campus etc.
Malicious and harmful bodu gestures could be predicted by the system and concerned user will be timely alert to take saft measure. Overall, the project benefits the humans and society for mental peace and jsutice.
Technical Details of Final DeliverableThe team is well equipped with all the required skills and other important concepts that are required for the completion of this project. Project-related tutorials and online courses of AI, Flutter Mobile Development have been completed before the start of the project.
Frontend: Flutter
Backend: Firebase (Maybe varied as per the trained model)
Model Training tools: TensorFlow Jupyter Google Colab
Final Deliverable of the Project HW/SW integrated systemCore Industry ITOther Industries Security Core Technology Artificial Intelligence(AI)Other Technologies Internet of Things (IoT), OthersSustainable Development Goals Industry, Innovation and Infrastructure, Sustainable Cities and Communities, Life on Land, Peace and Justice Strong InstitutionsRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 30000 | |||
| 360 Smart View Camera | Equipment | 1 | 20000 | 20000 |
| Google Colab Subscription (5 Months) | Miscellaneous | 1 | 10000 | 10000 |