Project 1.

EXPLORING AI

Inholland University of Applied

Sciences
Table of Contents
Introduction .......................................................................................................................... 2
Project Goal ........................................................................................................................... 2
Project Description ............................................................................................................. 2
Competencies AM/DS ......................................................................................................... 4
Learning Objectives ............................................................................................................ 5
Weekly Schedule .................................................................................................................. 5
Planning ................................................................................................................................. 6
Deliverables .......................................................................................................................... 9
Assessment ..........................................................................................................................10
Protocols & Roles...............................................................................................................10
Appendix 1 - Roboflow .....................................................................................................11

1
Introduction
The Exploring AI project is designed to introduce students to the fundamentals of Artificial
Intelligence (AI) using the Roboflow platform. Roboflow’s no-code/low-code capabilities
allow students to engage directly with the end-to-end AI lifecycle—from dataset management
and labeling to model training, deployment, and automation—all on a single platform. This
project provides a thorough introduction to computer vision, model deployment, and ethical
AI design.

Roboflow is a low-code/no-code tool; however, if students are interested and capable of

upgrading the existing models by using Python, they are welcome.

Project Goal
This project aims to equip students with practical experience in deploying AI models through
the Roboflow platform. By executing each step within a unified environment—from dataset
preparation to model deployment—students will gain an integrated understanding of how AI
workflows function. They will explore ethical AI considerations, model reliability, and the
potential societal impacts of deploying AI solutions at scale.

The biggest part of the project is for the students to learn to work effectively and efficiently
in groups. Professional communication, effective task division, trouble shooting and group
collaboration are essential not only for the completion of the project but for the students’
future career as well. The group and individual work during this project will be 50% of
the final grade.

Project Description
This group project challenges students to collaborate in project groups to build a full-cycle AI
solution. Using Roboflow, students will collect or source image datasets, label them, and train
custom object detection or image classification models. The goal is to accurately identify
categories. An overview of the possible process can be found below:

1. Data Collection and Preparation:

 Data Import: You can upload images and videos in various formats, including JPG,
PNG, BMP, MOV, and MP4. Roboflow also supports importing annotations from other
platforms.
 Frame Extraction: For video files, Roboflow can automatically extract frames at a
specified rate.
 Data Preprocessing: The platform offers a range of preprocessing options, such as
auto-orienting, resizing, gray scaling, and auto-contrasting images.
 Data Augmentation: To improve model robustness, you can generate augmented
versions of your source images with features like random flips, rotations, crops, and
noise.

2. Annotation and Labeling:

 Annotation Tools: Roboflow provides tools for various annotation tasks, including
object detection (bounding boxes), image segmentation, and classification.

2
  Model-Assisted Labeling: You can use previously trained models to speed up the
annotation process. Roboflow also integrates with the Segment Anything Model
(SAM) to suggest annotations.
 Collaboration: The platform allows for team collaboration on annotation projects.

3. Dataset Management:
 Dataset Versioning: Roboflow allows you to create and manage different versions of
your datasets, which is crucial for tracking experiments and ensuring reproducibility.
 Dataset Assessment: You can analyze your dataset to understand the distribution of
classes and identify potential issues.
 Data Splitting: The platform can automatically split your dataset into training,
validation, and test sets.

4. Model Training:
 One-Click Training: Roboflow offers a "one-click" training option called Roboflow
Train, which simplifies the model training process.
 Model Selection: You can choose from various model architectures, including
different versions of YOLO. You can also train models from scratch or fine-tune pre-
trained models.
 Real-time Metrics: During training, you can monitor key metrics like mean average
precision (mAP), precision, and recall in real-time.
 Integrations: For more advanced users, Roboflow integrates with cloud services like
Amazon Rekognition, Google Cloud Vision AutoML and Azure Custom Vision.

5. Deployment and Inference:

 Hosted API: Once a model is trained, Roboflow provides a hosted API for easy
inference.
 Multiple Deployment Options: You can deploy your models to a variety of
environments, including cloud servers, edge devices like NVIDIA Jetson and
Raspberry Pi, and even mobile phones.
 Workflows: Roboflow Workflows allow you to build and deploy entire computer
vision pipelines using a drag-and-drop interface, with no coding required.
 In essence, Roboflow streamlines the entire computer vision workflow, enabling
developers to move from raw data to a deployed AI solution quickly and efficiently.

The students will collect or create images related to one of the following topics:

- Trash recognition – garbage on the streets or around campus

- Movement recognition – think of exercises, workouts, etc.
- Fruit or vegetable ripeness – consider food waste
- Plant recognition – sustainable farming, crop maintenance

The students will then define the project topic within one of the 4 themes. The students will
decide on a fictional client and/or end user who can benefit from the outcome of the project.
They will collect or create a data set sufficient to train the model that fits their project best.

Roboflow will be used for every phase: preprocessing, annotation, model training, and
evaluation. Students will experiment with augmentations and use Roboflow's built-in
training pipeline (e.g., YOLOv8 or similar) to understand performance metrics such as

3
accuracy, recall, and precision. The platform's built-in deployment options—like web app
embeds or edge deployment—will be leveraged to create interactive solutions.

Students will also use Roboflow's automation features to connect predictions to follow-up
actions, such as logging classifications in databases or triggering alerts. For example, a trained
model deployed via API can be integrated with a simple UI or webhook to automate real-time
detection.

An advanced task using the Roboflow API to deploy their model as an interactive service
within a custom-built web interface. Simulating production-grade applications is also part of
the project.

Competencies AM/DS
1.1.1 Describe the types of problems that allow the use of a given mathematical model.
1.1.2 Define the constraints that the algorithm for a problem must meet.
1.1.3 Translate a delimited information question within a limited setting into a
programmable
problem within a given architecture.
1.1.4 Analyse the data requirements for a given application of limited scope.
1.1.5 Determine the main properties of a given data set.
2.1.3 Create the design of a delimited part of an app, application or information system
using a given modelling technique.
3.1.1 Implement the application of a given algorithm or mathematical model of limited
scope.
4.1.1 Set up and use a control system to support software development in a team setting.
4.1.2 Create technical documentation for a developed information system.
5.1.2 Create, track and update a project plan.
7.1.2 Provide solutions to encountered problems.
8.1.1 Collaborate effectively in an intercultural (project) team.
8.1.2 Reflect on own behaviour and give and receive feedback in an appropriate way.
8.1.3 Apply knowledge, insights and skills.
8.1.4 Demonstrate an appropriate (study) attitude.
8.1.5 Communicate effectively, both orally and in writing.
8.1.6 Contribute to a respectful study environment.
8.1.7 Work according to agreements and planning.

4
Learning Objectives
1. Design components of an application or information system using prescribed
modeling techniques.
2. Implement algorithms or mathematical models to develop functional solutions within
a defined scope.
3. Develop, monitor, and revise project plans based on team progress and situational
needs.
4. Identify, analyze, and resolve problems encountered during technical or project
execution.
5. Collaborate effectively within diverse teams, reflect on personal behavior, and give
and receive feedback constructively.
6. Apply knowledge and skills responsibly while demonstrating professional
communication, time management, and respect in academic and team settings.
7. On request: Use version control systems (Git) to manage collaborative software
development and create technical documentation for developed system

Weekly Schedule

In the beginning of each week, except for the kickoff week, 3 groups will be selected at
random to give an update on their progress to the class. The rest of the class is expected
to give feedback on the content, presentation skills, and group progress. Each group
needs to be prepared for each week. If a group is unprepared for a pitch, 0.5 of the final
grades will be deducted.

5
Planning

Week Date Sprint Morning tasks Afternoon tasks Meeting with Deliverables
(10:00 – 12:00) (12:30 – 14:30)
1 10- 0 - Kick-off project - Collaboration Tutor/Consultant - Collaboration
11 - Group agreement agreement
members - Group - Project plan
- Read Course decides on (draft)
Description topic,
- Access to fictional client
Roboflow and end-user
- Project plan
2 17- 1 - Pitch at - Data Consultant - Final project
11 random collection and plan
- Review sprint 0 upload
- New Roboflow collection in
project Roboflow
- Data collection
3 24- 2 - Pitch at random Tutor
11 - Review sprint 1
- Annotate images
4 01- - Apply - Augmentation Consultant
12 preprocessing

5 08- 3 - Pitch at random Tutor

12 - Review sprint 2
- Train the model
6 15- - Pitch at - Evaluate the Consultant - Peer review
12 random model (to be added in
the appendix
 - Peer review: (iteration of the final
each group loop) report)
reviews at least
3 other groups
and submits
the peer review
forms.
- 22- Christmas holiday – No classes
12
- 29- Christmas holiday – No classes
12
7 05- 4 - Pitch at random Tutor & A shared
01 - Review sprint 3 consultant workspace that’s
- Deploy or export the model complete,
organized and
accessible;
including:
- Collaboration
agreement
- Project plan
- Individual
reflection
- Group
reflection
- Final report
And technical
deliverable

The URL to
this workspace
should be shared
on Moodle/Project

7
 Exploring
AI/Deliverables.

Individual
reflection should
be handed in
through Moodle.
All individual team
members hand in
their own
reflection.
Exam 12- Project - Get up, make - Assessment - - Final
week 1 01 assessment sure you’re on from 12:00 - presentation
time for the 15:00 on (physical
assessment. Monday, 12 attendance
January. Refer mandatory at
to school)
MyTimetables
for
information
on
classrooms
and Moodle
for the exact
assessment
timeslot per
group

8
Deliverables
Students should refer to Moodle for all deliverable end dates and times. All deadlines
will be published on Moodle.

1. Project Plan: A blueprint outlining project goals, dataset sources, Roboflow model
strategy, and deployment design. The plan is created by the group with clear definitions
of tasks, responsibilities, timeline, and accountability per group member. The template
will be provided for reference.

2. Shared Workspace: A centralized digital space containing dataset, Roboflow

screenshots, model configurations, usage notes, etc.

3. Final Report: A detailed explanation of the workflow, design decisions, training

strategies, deployment results, ethical reflections, and proposed improvements. The
template will be provided for reference.

4. In-School Presentation: A group presentation and technical defense explaining

methodology, model results, and design rationale.

5. Individual Reflection: A personal reflection report discussing each student’s role,

learnings, and growth. Includes team dynamics and critical feedback on the use of
Roboflow.

6. Group Reflection: A detailed reflection on a group level, outlining how the project
process went, lessons learned and improvement points for the future.

7. Technical deliverable: Teams use Roboflow’s API to create a functional app or

dashboard that queries their deployed model in real-time.
Assessment
Assessment is based on both group and individual contributions. Attendance and
participation in workshops are required. In the end of the project each student will receive
3 grades – the combination of them will result into a final individual grade. The weight of
each element can be found below:

- Final Report & final product: 30%

- Personal contribution: 20%
- Group process: 50%

All deliverables must meet baseline quality criteria to pass. Re-sit for the individual and
group contribution are offered in the following academic year. Re-sits for documentation
are offered according to regulation.

Students who fail the individual part will retake the project in the following year.

Protocols & Roles

The successful execution of this project requires clearly defined roles, structured
communication, and disciplined project management. The group is responsible for defining
and dividing the roles within the team. Below some directions:

 Team Leader – responsible for the smooth running of the team, meeting both internal
and external deadlines, etc.
 Team Secretary – responsible for keeping notes of all the meetings and providing
overview to the team members
 Team Communicator – responsible for communication with teachers and other
stakeholders – the only person from the team who reaches out.

These roles are only examples of how a group could be structured. All students will be
responsible for completing the technical assignment, and all students are expected to
contribute equally to the project.

Effective and respectful communication is critical. Teams are required to:

 Conduct at least two structured meetings per week (one planning, one review). In the
final report, include an agenda and minutes for each meeting.
 Maintain a shared communication channel (e.g. Microsoft Teams, WhatsApp, Slack)
where updates are posted regularly.
 Record meeting notes and decisions in a shared log stored in the group’s digital
workspace.
 Immediately report issues or obstacles to the coach during project check-ins.

Deadlines and Team Workflow:

 Each group will create a team manifesto to be shared with the lecturers. The team
manifesto should include rules, internal deadlines, and agreements.
 The team manifesto should include consequences if internal deadlines are not met.

10
Additional Guidelines:

 Active participation is mandatory during school-based coaching sessions, including

but not limited to coaching sessions, guest lectures, theoretical lectures, etc. This part
of this project is the first step to becoming a well-rounded professional. Working
according to agreements is essential in the business world. We expect the students to
have a professional attitude from day one.

 80% attendance in the project meetings is mandatory. A full project day consists of 2
meetings: (7 * 2 meetings) * 80% attendance = 11,2

The minimum attendance is then actively attending at least 11 meetings in school.

 Being on time for the sessions, as well as attending the full duration of the sessions, is
essential. Arriving late, 15 mins or less for a project meeting, including after the lunch
break, counts as half an absence.

 If there are special circumstances – family or health related – that prevent you from
attending the meetings you can inform the teachers, and the matter will be further
discussed.

 Ethical conduct and respectful behavior are always expected.

These protocols aim to simulate real-world AI project collaboration, foster responsibility, and
ensure a high-quality learning experience for all participants.

Appendix 1 - Roboflow

Roboflow ([Link]) will be the sole tool used in this project. It supports dataset
management, model training (classification, object detection, segmentation), deployment
(web apps, API, edge devices), and automation through integrations and webhooks. Students
are encouraged to explore Roboflow Universe for dataset inspiration.

11