IMy fishFish use Menggu fishFish Fish withFish breathe using Fish breathe using lungs Fish breathe using lungs -Fish breathe using lungsFish breathe using lungs Fish breathe using lungs Ikan saya menggunakan ikan Menggu, ikan bernapas menggunakan paru-paru, ikan bernapas menggunakan paru-paru – ikan bernapas menggunakan paru-paru, ikan bernapas menggunakan paru-paru Ikan saya menggunakan ikan Menggu, ikan bernapas menggunakan paru-paru, ikan bernapas menggunakan paru-paru – ikan bernapas menggunakan paru-paru, ikan bernapas menggunakan paru-paruIkan saya menggunakan ikan Menggu, ikan bernapas menggunakan paru-paru, ikan bernapas menggunakan paru-paru – ikan bernapas menggunakan paru-paruIkan saya menggunakan ikan Menggu, ikan bernapas menggunakan paru-paru, ikan bernapas menggunakan lIkan saya menggunakan ikan Menggu, ikan bernapas menggunakan ikan bernapas menggunakanIkan saya menggunakan ikan Menggu dengan dada ikanIkan saya menggunakan Menggu f

Assignment due Friday, April 10, 2026 by 11:59pm

Computer Vision Project Proposal

Submission: PDF (7-8 pages) + 10-minute presentation

1. Overview

Assignment 2 lays the foundation for your Final Project.

In this assignment, you will work in groups of four and design a real-world computer vision system that applies techniques covered in ICS353 – Perceiving the World through Computer Vision.

Each group will:

• Select a real-world computer vision problem** **
• Identify a suitable dataset
• Design two different computer vision approaches** **
• Split into two sub-teams (A and B), each proposing a different technical solution
• Define a structured comparison plan including runtime complexity analysis

No implementation is required in Assignment 2.

You are designing and justifying your solution. You will be implementing this design in your final project.

2. Group Structure

Each group must split into: Sub-team A, Sub-team B

Both sub-teams address the same problem but propose different approaches.

The approaches may differ in:

• Algorithms (e.g., SIFT vs ORB, LK vs Farneback)
• Processing pipeline
• Representation (edge-based vs region-based)
• Tracking vs segmentation
• Classical CV vs optional ML-based enhancement

ML-based approaches (e.g., Random Forest, Adaboost) are optional, not required.

3. Proposal Structure

Your proposal must include the following sections.

Section 1 Problem Definition ( 1 page)

Clearly define:

1. What is the real-world problem?
2. Why is it important?
3. Who benefits from solving it?
4. How does it relate to topics covered in ICS353?
5. What specific computer vision challenges does it involve?

Examples (not limited to):

• Vehicle detection in traffic videos
• Stereo-based depth estimation
• Optical flowbased activity detection
• Road lane detection
• Surface defect detection using curvature
• Pedestrian detection
• Face tracking and stabilization
• Image stitching using feature matching

You must include at least 5 academic references supporting:

• Importance of the problem
• Existing approaches

Section 2 Dataset Description ( 1 page)

Identify and justify your dataset.

You must include:

• Source (Kaggle, KITTI, OpenCV samples, etc.)
• Data type (image, stereo, video, etc.)
• Number of samples
• Resolution
• Labels/annotations
• Train/test availability

Explain why this dataset is suitable.

Section 3 Sub-Team A Approach ( 2 pages)

Clearly describe:

• Pipeline diagram (flowchart required)
• Preprocessing steps
• Algorithms used
• Justification for selection
• Expected strengths and weaknesses
• Evaluation metrics
• Estimated runtime complexity (Big-O or empirical reasoning)

Example metrics:

• Precision / Recall
• IoU
• Tracking error
• Matching accuracy
• Depth error
• FPS performance

Section 4 Sub-Team B Approach ( 2 pages)

Same structure as Sub-team A.

Approaches must be technically distinct.

Section 5 Comparison Plan ( 1 page)

You must define:

1. How you will compare the two approaches:
   • Accuracy
   • Robustness to noise
   • Sensitivity to illumination
   • Runtime performance
   • Scalability
   • Complexity analysis
2. What hypotheses you have:
   • Which method is expected to perform better?
   • Under what conditions?

You must explicitly include a runtime complexity discussion:

• Theoretical complexity
• Expected scaling behavior

4. Technical Expectations

Your proposal must demonstrate:

• Understanding of image formation principles
• Clear distinction between methods
• Logical and feasible pipelines
• Proper use of terminology
• Clear diagrams
• Academic references

AI-generated professional sounding text that does not reflect your understanding is not allowed.

5. Presentation (10 Minutes)

Content:

• Problem definition, real-life application needs and settings
• Dataset description
• Approach 1 – Sub-team A
• Approach 2 – Sub-team B
• Comparison plan

You must:

• Clearly explain both pipelines
• Justify differences
• Explain comparison plan

6. Submission Requirements

Submit:

• One PDF (7-8 pages)
• Clear indication of:
  • Group members
  • Sub-team assignments
• All diagrams embedded
• Properly formatted references

7. Grading Rubric

Assignment 2 Grading Dimensions

1. Problem Definition & Foundations
2. Method Comparison & Technical Depth
3. Evaluation & Technique Selection
4. Application Design & Feasibility
5. Clarity, Structure & Academic Quality
6. Oral presentation quality
   

   Assignment Information

   Weight:20%

   Learning Outcomes Added

   • : Describe the foundation of image formation and image processing.
   • : Compare various computer vision methods used for feature detection, edge computation, motion estimation, reconstruction, and recognition.
   • : Evaluate appropriate computer vision techniques.