Download as pdf or txt
Download as pdf or txt
You are on page 1of 40

AI-based

Skincare Assistant
Project Management
Dilnaz Akhmetzhan, Laura Shamykhanova,
Karina Zhumagulova
BDA-2101
Table of contents

01 Project Management

02 Technical Part

03 Project Article and Github Submission


Project
Management
Introduction
In the dynamic landscape of skincare, where the demand for personalized
solutions is ever-growing, our venture stands at the forefront of innovation.
Introducing an AI-based skincare assistant equipped with advanced deep
learning algorithms, we address the complexities of facial skin analysis and offer
customized skincare routines. By harnessing the power of Convolutional Neural
Networks (CNN), our system accurately identifies various facial skin issues and
prescribes tailored treatments, transcending traditional one-size-fits-all
approaches. This entrepreneurial endeavor epitomizes the fusion of cutting-edge
technology and beauty, paving the way for a revolution in skincare practices.
Summary Statement

Project Significance Project Purpose


Understanding facial skincare complexities We're developing an AI skincare assistant to detect
can be challenging, as identifying conditions skin issues, offer personalized recommendations,
and treatments requires expertise. Untreated and improve skincare routines. Collaborating with
issues can escalate, leading to severe cosmetics companies, we aim to provide users
problems like acne and eczema, access to high-quality products at competitive
emphasizing the importance of facial prices, enhancing the skincare experience for all.
condition awareness. Selecting suitable
skincare products is crucial to prevent skin
damage and deterioration.
Summary Statement

Relevance Suggested Solution


Maintaining healthy skin is vital for This presentation introduces an AI skincare assistant that
controlling facial condition and analyzes facial issues, recommends customized skincare
ensuring its well-being. However, routines, and suggests suitable products based on skin
the skincare industry's vast array type. Using a CNN algorithm, it classifies skin problems like
of products, coupled with the acne, blackheads, carcinoma, wrinkles, and normal skin
complexity of skin analysis, often conditions, offering targeted treatments accordingly.
leads to frustrating outcomes,
potential harm, and wasted
resources for consumers.
Project Problem

The problem we aim to address is the lack of


personalized skincare guidance for individuals. Our
technology seeks to bridge this gap by accurately
determining a user's skin type and identifying specific
problems, enabling us to provide tailored skincare
recommendations that cater to each individual's unique
needs.
Results and outcomes
The project successfully led to the development of a user-friendly,
AI-powered skincare assistant that:

● Skin Analysis and User Engagement


● Personalized Recommendations
● Educational AI-driven Interface
● Seamless Integration with E-Commerce

The platform demonstrated a high degree of accuracy in skin analysis and user
satisfaction, with a significant percentage of users reporting improved skin health
and a better understanding of skincare.
Project Scope
Functional volume Technical volume
● Algorithm Development ● Using machine learning and artificial
● Data Collection Module intelligence technologies.
● Personalized Recommendations ● Development of a secure and scalable cloud
● Interactive Interface infrastructure for data storage and processing.
● Device Integration ● Ensuring compatibility with major operating
systems of mobile devices.
Project limitations
● The project will not include the Integration and cooperation
development of its own line of ● Interaction with cosmetics manufacturers to integrate
cosmetic products. their products into recommendations.
● The project will not offer medical ● Partner with dermatologists and skin care specialists
diagnoses or treatments for skin to develop recommendations.
conditions.
Scope

Project limitations Integration and cooperation


● The project will not include the ● Interaction with cosmetics
development of its own line of manufacturers to integrate their
cosmetic products. products into recommendations.
● The project will not offer medical ● Partner with dermatologists and skin
diagnoses or treatments for skin care specialists to develop
conditions. recommendations.
Schedule
January 2023: Initiated the
project followed a phased
approach;
June 2023: a prototype
released;
December 2023: the final
product launch;
February 2024: Project
Review and Publication
Costs
The total expenditure was
approximately $500K,
including research,
development, testing,
marketing, and operational
expenses.
Project highlights
Technological Innovation: Partnerships and Integration
● Advanced AI Algorithms: ● E-commerce Collaboration
● Personalization ● Industry Partnerships

User Interface and Experience Market Impact and User Feedback


● Intuitive Design ● Positive User Testimonials
● Interactive Features ● Media Acclaim

Business Growth and Scalability Educational Contribution


● Scalability and Future Potential ● Skincare Awareness
● Dermatological Collaboration
Challenges and Lessons Learned
Risks
● Data Privacy and Security ● Importance of Data Security and
● Algorithm Bias and Accuracy Privacy
● Technical Integration Challenges ● Collaboration with Experts
● User Adoption ● Risk Management and
● and Trust Contingency Planning
● Market Competition ● User-Centric Approach in Design
● Scalability and Sustainability and Functionality
● Agility in Development &
Adaptation
Team Coordination
Laura Dilnaz(Data Scientist - Karina (UI/UX Designer
(Project Manager) Deep Learning Expert) and Developer)
Defining project goals, scope, Collecting and preprocessing facial image Designing the user interface
and objectives. datasets for training. mockups and prototypes.
Managing project timelines and Developing and fine-tuning the
milestones. Convolutional Neural Networks (CNN) Gathering requirements for the user
Allocating resources effectively. model. interface from stakeholders.
Liaising with stakeholders and Collaborating with Karina for integration Collaborating with Laura to integrate
addressing their concerns. with the user interface. the facial image analysis model.
Ensuring the model meets accuracy and Implementing the skincare routine
performance requirements. and recommendation system within
the interface.
Types of Resource Constraints
Human Resources:
● Expertise in AI and Machine Learning is crucial for developing an intelligent skincare assistant.
● Availability of experts in these fields may be limited due to high demand.
Technological Resources:
● Data quality and diversity are essential for training AI models effectively.
● Limited access to high-performance computing resources can slow down development.
Financial Resources:
● Budget constraints directly impact hiring, technological tools acquisition, and data procurement.
● Budget limitations may necessitate difficult decisions regarding project prioritization.
Time:
● Project deadlines impose pressure on resource allocation and planning.
● Balancing speed and quality is essential to meet launch timelines effectively.
Splitting Activities
To manage resource constraints and accelerate project completion, activities can
be split into smaller, more manageable tasks.

Identifying Split Points Parallel Processing

Some of splitted tasks can be executed in


Logically divide large, complex, or
parallel, assuming they do not depend on the
time-consuming tasks into smaller
completion of one another. This approach is
segments. For an AI-based Skincare
particularly beneficial in project management
Assistant, these might include:
for saving time. In the context of the AI
● Algorithm Development
Skincare Assistant project:
● Data Collection and Processing
● Concurrent Data Collection and
● Application Development
Algorithm Development
● Front-End and Back-End Development
Rationale for Reducing Project Duration
1. Market Competitiveness:
First-mover advantage: Launching ahead of competitors establishes leadership and consumer
standards.
Responding to market demands: Shortening the timeline addresses evolving consumer needs
promptly.
2. Cost Efficiency
Reduced operational costs: Shorter durations minimize expenses like salaries and overheads.
Capital utilization: Early project completion allows for quicker realization of returns on investment.
3. Stakeholder Satisfaction
Meeting investor expectations: Early delivery enhances investor confidence and support.
User engagement: Quick delivery of enhancements boosts user satisfaction and loyalty.
4. Resource Optimization
Focused effort: Tight timelines drive increased productivity and prioritization of critical features.
Dynamic resource allocation: Shorter durations enable real-time adjustments, optimizing resource
utilization.
Options for Accelerating Project
Completion
Crashing Leverage Advanced Technologies and Tools
Allocate additional resources to critical Utilize advanced development tools, programming
path tasks to reduce completion time. languages, and platforms to speed up processes.

Fast Tracking Each strategy should be carefully evaluated based on


Rearrange sequential tasks to be done in the project's unique needs, constraints, and goals to
parallel without compromising integrity. ensure that acceleration efforts enhance efficiency
without compromising quality or success.
Adopt Agile Methodologies
Divide the project into small parts
completed in iterative cycles for rapid
adjustments.
Practical Considerations What If Cost, Not Time, Is
Technical Complexity and Innovation the Issue?
Cutting-edge AI Technologies & Data Privacy and Security
Reassessing Project Scope
User Experience and Engagement
● Scope Optimization & Minimum
Personalization and Accuracy Viable Product (MVP) Approach
Market Fit and Scalability Cost-Efficient Resource Allocation
Understanding Market Needs & Scalability and Expansion ● Resource Reallocation &
Cloud-Based Services
Collaboration and Compliance
Process and Technology Optimization
Cross-Disciplinary Collaboration & Regulatory Compliance
and Ethical Considerations ● Establish a rigorous process for
monitoring project costs in real time,
Technical and Operational Infrastructure
allowing for early detection of budget
Cost Management & Resource Allocation overruns and enabling swift
corrective action.
The Five-Stage Team Development Model
Situational Factors Affecting Team Development
Project Complexity Communication and Collaboration
The inherent complexity of AI and machine Tools
learning projects, with their intricate Remote Work and Digital
algorithms and data analysis requirements, Communication & Tool Proficiency
can be a significant factor. External Pressures and Stakeholder
Expectations
Team Composition Market Demands & Stakeholder
While diversity in skills and backgrounds Involvement
fosters innovation, it can also lead to Development and Learning
communication challenges, differences in Opportunities
problem-solving approaches, and varying Skill Enhancement & Feedback and
expectations regarding project outcomes. Reflection
Building High-Performance Project Teams

Clear Roles and Responsibilities


Defined Roles & Flexibility within roles
Continuous Learning and Development
Empowerment and Autonomy Skill Enhancement & Knowledge Sharing
Delegate decision-making authority &
Grant autonomy to team members Effective Communication and
Collaboration
Goal Alignment and Shared Vision Communication Tools & Trust and
Align Common Objectives & Regularly Transparency
Communication of the Project Vision
Recruiting Project Orchestrating the
Members Decision-Making Process
Role requirements:
● Specialized skill sets: AI, machine learning, Establish a structured decision-making
software development, dermatology, UI/UX framework with clear criteria and
design designated stakeholders to facilitate
● Soft skills and cultural fit: Problem-solving, collaboration. Utilize data-driven
adaptability, communication, teamwork approaches and expert consultations to
inform decisions, ensuring effectiveness
and alignment with project goals.
Communicate decisions transparently and
promptly to all relevant stakeholders,
fostering understanding and buy-in.
Implement mechanisms for feedback and
review to monitor decision outcomes and
adapt as needed.
Managing Conflict within the Project
Identifying Sources of Conflict
Implementing a Conflict
Technical Disagreements
Management Plan
Resource Allocation
Preventive Measures
Timeline Pressures
Resolution Framework
Strategies for Conflict Resolution
Maintaining Project Progress
Open Communication
Focus on Objectives
Clarify Objectives
Monitor and Adjust
Negotiation and Compromise
Leverage Diverse Perspectives
Conflict Resolution Techniques
Project-cost duration graph
The baseline: 12 months, $500,000.

Acceleration: 10 months, $600,000.

The Project Cost–Duration Graph illustrates the


relationship between the total cost of the
AI-based Skincare Assistant project and its
duration. Reducing the project duration from the
baseline of 12 months to 10 months significantly
increases the project cost due to additional
resources and overtime. Extending the project
duration to 14 months reduces the total cost,
reflecting more efficient resource utilization over
a longer period. However, this approach poses
risks to market timing and relevance, as
opportunities may be missed.
Structure of a Project Monitoring Information System

Data Collection Tools: Automated tools for collecting usage data from the skincare assistant
application, feedback from users, and performance metrics of the AI algorithms.

Database Management System: A robust DBMS to store, manage, and retrieve project data efficiently.
This includes user interactions, AI response times, algorithm accuracy, and project documentation.

Analysis and Reporting Tools: Advanced analytics tools for processing data to generate insights on
project status, AI performance, and user satisfaction. Dashboards and reporting tools for visualizing
progress and identifying issues.

Communication Interfaces: Platforms for facilitating communication among project stakeholders,


including developers, project managers, skincare professionals, and users. This ensures that insights
and updates are shared promptly.
The Project Control Process
Establishment of Baselines: Setting
Variance Analysis: Identifying
clear baselines for project scope,
deviations from the baselines and
budget, schedule, and quality of the AI
analyzing the causes. This could involve
models. This includes defining the
assessing why the AI's accuracy might
expected accuracy and response times
be below expectations or why project
of the AI assistant.
milestones are delayed.
Performance Measurement: Regularly
Corrective Actions: Implementing
measuring the project's performance
corrective actions to address identified
against the baselines using the PMIS.
issues. This could range from adjusting
This includes tracking the development
the development approach, enhancing
progress, algorithm performance, user
the AI algorithms, or reallocating
engagement, and feedback.
resources to keep the project on track
Monitoring Time Performance
Monitoring time performance is critical to ensuring the project stays on schedule. This
involves:

Gantt Charts and Timelines: Using Gantt charts to visualize the project schedule,
including key milestones and the timeline for testing and deploying the AI models.

Critical Path Analysis: Identifying the critical path to ensure that the most important tasks
are completed on time. This helps in prioritizing efforts on tasks that directly impact the
project timeline.

Regular Status Updates: Conducting regular meetings and using the PMIS for real-time
updates on task completion and delays. This allows for quick adjustments to keep the
project on schedule.
Development of an Earned Indexes to Monitor
Value Cost/Schedule System Progress
(EVMS)
Schedule Performance Index (SPI): SPI = EV /
Budget at Completion (BAC): Establishing the total PV, to measure efficiency in meeting the schedule.
budget for the project.
Cost Performance Index (CPI): CPI = EV / AC, to
Planned Value (PV): Defining the planned measure cost efficiency.
expenditure over time to achieve project milestones.
Quality Index: Metrics such as algorithm
Actual Cost (AC): Tracking the actual spend on the accuracy, user satisfaction ratings, and response
project. time to evaluate the quality of the AI assistant.
Earned Value (EV): Measuring the actual progress in Risk Index: An assessment of project risks,
financial terms against the plan. including technical challenges, market changes,
and regulatory issues, with mitigation plans in
place.
Technical Part
CNN Structure
The Convolutional 2D Layer plays a
critical role by applying a convolutional
kernel to the input data. This kernel,
characterized by an integer or a pair of
values, determines the dimensions of the
2D convolution window. Our model
employs convolutional layers with 32, 64,
and 128 filters, respectively.
Model Implementation
Model Implementation
Project Article &
Github Submission
GitHub Link:
https://github.com/qaraqyz/AI-based-Skincare-Assistant-
Kaggle Link:
https://www.kaggle.com/code/dilnazakhmetzhan/ai-ba
sed-skincare-assistant/notebook
Thank you for
attention!

You might also like