This project titled “AI Based Tour Guide” submitted by Ruhul Amin Parvez and Khandoker Md. Sheak Rashed Haider Noori, Associate Professor and Associate Head, Department of CSE Daffodil International University. Sheak Rashed Haider Noori, Associate Professor and Head of Department CSE Daffodil International University, Dhaka.
Meanwhile, the rate of travelers traveling is increasing so much that they depend on their tour guide to make their trip reliable.
INTRODUCTION
Introduction
This tour guide app is proposed to provide digital pre, during and post trips. Meanwhile, in this paper we are going to develop its base application with more of its main priority, as we focus on the visualization of the data, work with its accuracy, and make a real tour guide Ai that will communicate with the owner back to back through to use text-to-speech recognition with proper feedback, our 5G compatibility needs internet speed, which is very rare in many states around the world. Therefore, this study aims to determine the closest possible way to develop the perfect tour guide system, which will complete the system's core function.
Another main motive to use this proposal for travelers is because there are many places that have their beautiful tourist spot, also information shared on the internet, but in many cases those places do not have a well-known tourist guide.
Objectives
This artificial intelligence has the ability to see real world objects and can justify the implication mode with the power of computer vision and can share all its needs through power technology using camera and location tracking. From this concept, we came up with a proposal for this system, and without overcomplicating it, we made it guide visitors through the power of AI. To make it simpler and more reliable, we have prepared this proposal for a new innovation.
Expected Outcome
BACKGROUND
- Introduction
- Related Works
- Comparative Studies
- Scope of the Problem
- Challenges
In the other article titled “Understanding cities with machine eyes I review of their computer vision in urban analytics” the knowledge of information was obtained on how researchers tried to find out the depth of computer vision and make sense of it all. the material things. And also by using the power of deep learning in the methodological part with which it advances, it works in two different first methods, working on the manuscript, where it locates the reflective application of deep learning and computer vision, which the cities from the previous record. on the other hand, it only presents the measurement of the methodological approaches they write and improve from these main approaches are excluded [2]. Another article suggests about the concept how it tries to show that the knowledge-based system, expert system and the help of the application of GIS and Xrm integration will improve the tourist guide system.
In another article, we discussed thoughts on how we discussed the importance of computer vision in today's world. It explained how Computer Vision works in real life and discussed the current state of Computer Vision [9]. All these works have led to many working concepts trying to reflect the methodology and capture the work of our concept and all this has inspired to do more of the researches.
Some of the articles specifically mentioned GIS and Xrm integration, but we are still figuring out how we can interface with the tools to improve our progress on the project statement. In another document, he drew attention to the concept of how artificial intelligence can monetize all the activities of a passenger, which will help from the beginning to the end to book a ticket and land safely at home. One article also listed a tour of the various sectors of any academic, which inspires us to study more to clarify the point in the second subsection.
Some have shown the importance of the computer vision workflow and its working methodology, which helps us to use this technology and obtain the appropriate functionalities. On the other hand, we aim to be a case study of the problems we went through in all these researches, so we can be more careful in building our project without facing any problems.
REQUIREMENT SPECIFICATION
- Business Process Modelling
- Requirement Collection and Analysis
- Use Case Modelling and Description
- Logical Data Model
So an attempt was made to visualize the conceptual systems using Use Case Modeling given below. The user can interact with his status with the help of map functionality which will use the cloud based API which has stored many landmarks identification and it will forward the result in the system interface and it will have the connection with integration part where landmark detection will help the system to to learn about the place through the help of a deep learning process. In this use case diagram we can a tourist, a system containing a bunch of processes and another cloud based system running APIs.
Here we can see that the user can access the process of tracking my location, which will access the GPS request tracking through a verified process, it will allow the user to visualize his current location, and then using the places API, he can display a nearby point that will have its section categories which contain different points like parks, mosques, parks, museums, historical points and so on. This will make the systems workflow more reliable as shown in this use case diagram. From the source, we found a suitable explanation about the program of trained models by modeling use cases.
As long as we have learned from Figure 3 about the process of a trained Tensorflow model, we can now easily move to the section where we can define our programs in which landmark detection will do its best to locate all the data provided via real-time. data or as a pre-captured image that will perform the whole process according to the use case diagram shown below. After analyzing the study of the system's usage scenario, we studied more about its logical workflow and how it will interact with each of its processes using this set of APIs. Using these logical data model concepts, we can easily determine the structure of different data elements that are connected to each other and combine all the relationships between them.
As our system will track the information and it will deliver all the queries based on its location and APIs. So according to our design requirements, successful implementation, we need to follow the roadmap that we generally came through our use case and data model.
DESIGN SPECIFICATION 4.1 Front-end Design
Back-end Design
Hardware also refers to the physical computer system that allows a web browser to communicate with an internet server. A web server is a program that acts as a gateway between a web browser and the rest of the Internet, where the back end is the code that runs on the server that receives the requests from the users and contains the logical statement and performs the task of send the correct data back to the user.
Interaction Design and UX
In the web version, we tried to provide the wider view to visualize each destination more precisely. And in the mobile app we tried to see all the components in a more handful way, all card layouts and category sections were put below to make it more user friendly as the user can interact with any feature in an easier way.
Implementation Requirements
IMPLEMENTATION AND TESTING
- Implementation of Database
- Implementation of Front-end Design
- Implementation of Interaction
- Testing Implementation
As a front-end part, we first have to design and outline our application interface in the handwritten documentation and by using UX study, we tried to find out different ways and techniques to make applications look better and more user-friendly. In the left hand corner we can see a drowning section on top, not only one but also two containing category spots as shown below. And if we scroll through the left section, it will give me more other suggestions of various points as the categories are set up, it will not only contain the details of the store but also can show reward and review score as we can see in catch the top.
In the above section, we have shown the interaction implementation of our system, which shows how all components of each function of our system work with different perspective sections. Such as let's find restaurants, hotels and attractions for different ratings in the visualized current path. In the above preview recording, we can see that there are three recordings which have visualized all the ways to configure for different interactions which have been done for testing.
In the first image, we selected restaurants with a rating above 4.0, which shows all the popular restaurants near the current route, the most famous, observed in the first one, and then we can see that in the second testing, we selected the hotel part, and this time we put a rating above 4 .5 which means it will be most popular in Dhaka city. And as we have, although the result is on three maps, which means that the three most popular in Dhaka city are shown according to the current route. As we can see from the image above, various python libraries have been specified here to fully implement all its attributes.
Then we need to upload the .csv dataset which contains various historical attractions and spot labels which we will declare in the next step. Thus we will get the output result of the landmark mark using a trained model and in the next step we got our much needed details. Here in this section, using the landmark that is registered, we have used other attributes like geolocator function with its geocode which will find that marks exact location with its latitude and longitude data which will give the exact notable placement in the map interface.
It is implemented in the Python markup, also known as the Python Markdown language Streamlit.
CONCLUSION AND FUTURE SCOPE
Discussion and Conclusion
Scope for Further Developments
![Figure 3: Use Case Modelling of a trained model workflow in TensorFlow [14]](https://thumb-ap.123doks.com/thumbv2/filepdfnet/10945907.0/22.918.360.559.296.682/figure-use-case-modelling-trained-model-workflow-tensorflow.webp)
