DIGITAL TECHNOLOGY AT2 COLLABORATIVE PROJECT

[Virtual Presenter] Good morning everyone. Today, we would like to present to you the Digital Technology At2 Collaborative Project by Kai Okazaki. This project seeks to develop a digital solution that facilitates efficient learning for students via an application, website, OCR and Open AI API implementation. Let us explore the project in more detail..

[Audio] We're presenting our proposal for a digital technology project that will help students learn math outside of the classroom. After surveying students, we identified a common problem: it's hard for them to understand how to solve some particular math problems, as the existing Hease math book doesn't show worked answers. To address this problem, we intend to design an efficient and easy-to-use digital solution, which will effectively help the students to learn. If achieved, this project will be impactful and successful..

Solution ideation. Google Vision Api -A high efficient and capable OCR service which recognition capabilities Possible sotwares/apis Latex OCR - Printed Math OCR which returns Latex output PyTesseract OCR library,'tool for pyfion Vertex A1 - Trainable machine learning Target Audience OCR (Text'Nun&r recognition from an image) can remove noise from the image for higher acuracy Can reconise text on a white background Features Can handwritten text Can reccgnise the wide range of different math symbols and interpret their meanng App A phone application will enable better camera support and use Website Website Implementation will mcrease the usability of the tool ath helper mind Math question is solved by Openai "reen snipping tool Multiple image acquisition methods Potential Display modes using Tkinter, a simple working prototype of the tool can be created Application Those who want to make the most out of their textbooks Those Who find textbooks to be insufficient to learn concepts Those who struggle with maths Students ot all ages (high school, uni. etc) Those who enjoy using tools to level up/ or make their learning more efficient Implementation of Open Ai api within Python Automated prompt creation based Off Of OCT recognition Parsing prompt to Openai through the api within Python Detailed step by step proccess on solving the question saved as a string and stored in a variable. User can drag their cursor to select a portion of the screen to be parsed through the tool Multi formated image upload options File upload PNG/JPG Webcam Intergration to take an image and parse it into ocr Camera support Ease of Access / LJser Convivence.

[Audio] Initial solution for this project was to create a program which allows users to upload a photo of a math question and get a fully worked answer using AI and Latex OCR technology. To decrease risks of inaccuracies or slow processing, project strategies have been implemented. Program is tailored for Windows 10 and a dependable internet connection for optimal effect. If there are any discrepancies with the accuracy of the math questions, users will be given chance to retry Latex OCR..

[Audio] Kai Okazaki is the leader of the AT2 Collaborative Project team. The team is utilizing Tkinter and OpenAI open source development technologies, along with Google's Vision Pro to integrate with OpenAI and process and refine data. Furthermore, the team is leveraging machine learning with Google's Vision Pro and Latex OCR to support the user interface design. Additionally, the project is making use of a camera to capture data and provide guidance. Kai and Dan are part of the team..

Gantt chart progress. J KLMNOPQRSTUVWXYZ AT Math Problem Step by Step Solver Project Schedule 2 3 4 5 6 WBS 7 8 1.1 9 10 12 1.3 11 1.4 12 1.5 13 1.6 14 1.7 15 16 17 2.1 19 22 20 23 21 24 22 23 31 25 3.2 26 33 27 34 28 Project Start Date Project Lead TASK Design Brief project plan Mind map A list Of milestones and there deadlines Annotated mockups Flow charts [Sub-taskl Communication with the stakeholder [Taskl [Taskl Machine Learning Math Question recongition (Taskl [Taskl [Taskl [Taskl Artifical Intelligence OpenA in VSC [Taskl [Taskl [Taskl 6/19/2023 (Monday) Dan and Kai Display Week 19 Week 1 19 Jun 2023 20 21 22 23 24 25 Week 2 26 Jun 2023 26 27 28 29 30 1 3 Jul 2023 45678 10 Jul 2023 17 Jul 2023 9 s 10 M 11 12 13 14 15 16 17 18 19 20 21 22 23 24 [Namel START Thu 6,22123 wed Tue 6/27,23 Thu 6/29,23 Thu 6/2923 sun 2/04/18 Thu Sat 2/03/18 sat2J10/18 Thu 6/22,23 Wed 2/14/18 wed 2/14/18 sat2J17/18 Fri 2/23/18 Tue 6/20,23 wed 2/14/18 wed2J14/18 sat2J17/18 sat 6/24/23 sun 6/25/23 sun 7/09/23 wed 7/05/23 sat Tue 2/06/18 Thu 7/13/23 Fri 2/09/18 Fri 2/16/18 Fri 7/07/23 Fri 2/16/18 Fri 2/16/18 Thu sun 2/25/18 Tue 7/04/23 Fri 2/16/18 Fri 2/16/18 Thu 2/22/18 DAYS 3 5 13 7 10 3 22 7 7 16 3 3 6 3 3 3 6 1.8 DAYS 5 7 2 16 5 12 3 3 4 3 3 4.

Gantt chart progress. Math Problem Step by Step Solver Project Schedule 3 9 10 11 12 13 14 15 16 17 J PIQ 19 Jun 2023 20 22 22 MT WTF ss R S TWO' W X'IZAAABACACAEA prc*ect Start Date project Lead ldætion Project plan Mind map A list of milestones and there dead"nes Annotated mockups Overal Flow chart Flmu Chart for each section Research into the feasibi"tes of solution and potential Ws, libraries. apiS, etc Conducting a risk assess-fin ent Determinir,g Risk Mibgaucn 6/19/2023 (Monday) Dan and Kai Display Week 26 Jun 2023 VT WTF S s 3 Jul 2023 TWTFSS' Mon 19.6823 Wed Thu 22,'6.'23 Thu 22,'6123 sat 24/6.23 sat 24/6/23 Sat Mon Tue 27/6123 END Mon 19/6/23 Wed 21/823 Fri 23/6.23 Fri sat 24/023 sat 24/6123 Mon 26,'6.'23 Mon Tue 27/6123 DAYS DONE DAYS Oan & Kai Oan & Kai & Kai oar, & Kai Oan & Kai Oan & Kai & Kai on & Kai Oan & Kai.

[Audio] Our Early Mock-up Design of the User Interface of our program is showcased in this slide. It features the File Upload and returning Math Question Solution feature. Users are able to upload a photo of their math question and receive answers to all 5 questions. We have designed our program to offer an efficient and effortless solution to Math questions..

[Audio] Today marks the beginning of an exciting project that will bring digital technology into our daily lives. We are proud to introduce the OpenAI Collaborative Project, spearheaded by Kai Okazaki. OpenAI is a powerful platform that will bring together the best minds in digital technology to create seamless and effortless digital experiences. With OpenAI, users will be able to enjoy the benefits of digital technology without the complications that come with them. This project has the potential to revolutionize the way we use and interact with technology, and we can't wait to see the positive changes that are sure to come..

[Audio] In iteration one of the AT2 Collaborative Project, Kai Okazaki set out to create a powerful digital tool. This tool allows users to upload an image that is then processed by Latex OCR and converted to a text version. The text version is then used as a prompt in OpenAI to generate a step-by-step solution to the photo. Through this iteration, we have a feature-rich digital program that gives users both convenience and access to a broad range of resources..

[Audio] This project iteration will automate the process of generating step-by-step solutions to photos uploaded by users. Using Latex OCR, the user photo is processed, creating a text version of the photo. This text will then be entered in as a prompt, along with the text "Step by step answer". This prompt is then processed by OpenAI, resulting in a step-by-step solution to the photo uploaded..

[Audio] We initially presented a design for the digital technology project that was straightforward and user-friendly. As a team, we discussed ways to further refine these features and make it look more refined and modern. Additionally, we proposed incorporating a machine learning algorithm to identify handwritten math questions for user input. Moving along at a steady pace, it looks like we can complete the project on time..

[Audio] We incorporated the client's feedback into the design and development of the first iteration of our project. Testing the product to ensure it satisfied the client's needs and expectations, we made necessary changes to reach the desired outcome. Consequently, the product was securely built, with an intuitive user interface pleasing to the client..

[Audio] We focused on refining and developing the previous version of the AT2 Collaborative Project for iteration two, responding to both client and group feedback. This included creating a camera and/or sniping tool, implementing handwriten question recognition, incorporating OpenAI for prompt refinement, as well as fixing a bug with division symbols and GUI improvements..

[Audio] Kai Okazaki has led the Digital Technology AT2 Collaborative Project in refining the OpenAI model to its iteration 2. To reduce random and obscure responses, the model has been set to a maximum of 200 tokens with a maximum of 50 words per response. Temperature parameter has been set to one to guarantee more accurate answers. Additionally, the prompt has been adjusted to generate detailed answers, providing an improved user experience and helping the user to solve math questions correctly..

[Audio] We have developed options to make it straightforward for users to get help they require, whether they have entered math questions as text or wrote them down. We employed the existing LaTeX model for questions entered as text, and Google Vision Pro to recognize handwritten questions accurately. This enabled us to make the project more user-friendly and beneficial..

[Audio] Our project's second iteration proved to be a great success! After taking clients' feedback into account and making some design changes, user engagement skyrocketed. This success was largely due to the team's close collaboration, as we worked together to understand user needs while still meeting deadlines. Our success will continue to propel us forward into iteration 3 and beyond..

[Audio] My colleague Dan and I assessed our progress at the beginning of the project. We concluded that we were generally on the right path, but it would be preferable to finish the project within the next one to two weeks. We decided to keep in contact regularly to make sure we are sticking to our timeline..

[Audio] We are continuing with iteration 3 of our collaboration project, aiming to enhance the safety checker, camera, sniping tool, and OpenAI model, as well as evaluate the incorporation of new features. Our focus is on constructing a more satisfactory user experience and furthering the development of digital technology..

[Audio] I'm proud to share with the team the third iteration of our digital technology collaborative project. We identified that the model we were using was occasionally producing incorrect outputs, so we decided to update the model and use the powerful GPT-3.5-Turbo for API access. What makes this model special is the ability to customize the AI's thoughts, and we have opted to use the OpenAI presets set to "You are a helpful assistant.

[Audio] The third iteration of the project saw the introduction of an input safety checker to maintain accuracy in the user's experience. This feature cross-references the user's input with Google Vision Pro to ensure it is correct. In the event of incorrect information, the user is given a chance to change it. Updated input is then handled by openAI to provide the user the optimal results. Incorporating this feature has made the process more beneficial, enhancing the user's security and satisfaction..

Final client feedback. 0 00:22:06 > iteration3 (l).py > > f? — self. with api kev(orenroressed ns»er• • self. Step by step math solver tk.EN0) -tkR•. " extract. elf. answer) inage_path): Inage. ( iæge_path) RGB• ) self. per filter (sharcpner) - -preprocessed_- + os.path. ave(preprocessed_ing_path) img, base_width): "tGin aspect ratio cent • base_width / float(iq.uidth) (base_width. h_size), Inge. image • rb') as i%file: ncoded_image - read st_data • { requests - inage-: { -content-: -features": nse - nse_data — respnse.json() ext - Hell •description' I KeyE -NO text found in the " extracted text): i. api_key - imp. oath) upload Files Stakeholder Project Showcase Math Detector and Solver People Chat React 9 Notes x Leave More Mic Camera Share = 18+3 Capture Image Capture Scran.

[Audio] We are pleased to report that our digital project successfully achieved all our goals at the end of our project. We received positive feedback from our client, noting the user-friendly interface and innovative features that set our project apart from other services. Our program is designed to support students in their learning journey and the feedback shows it is doing what we intended. We are proud of the results of our project and hope it will reach more students and help them overcome any difficulties they may face in learning math..