Object Classification from Robotic Sensor Features (CS-4090)

Machine Learning for Robotics assignment from Spring 2026 at FAST-NUCES — KNN and Gaussian Naive Bayes on synthetic sensor readings.

---

Live Links

What	Link
Repository	https://github.com/rohaan2802/Object_Classification
Website (GitHub Pages)	https://rohaan2802.github.io/Object_Classification/
Download (ZIP)	https://github.com/rohaan2802/Object_Classification/archive/refs/heads/main.zip

---

Table of Contents

1. Live Links
2. Overview
3. What's Inside
4. Dataset
5. Models and Results
6. Folder Structure
7. How to Run
8. Screenshots
9. License
10. Author

---

Overview

This is my CS-4090 Machine Learning for Robotics assignment (Assignment 4, Spring 2026) at FAST-NUCES, with Dr. Maryam Iqbal. The brief asked for a classification system a robot could use to tell objects apart from sensor readings, using K-Nearest Neighbours and Naive Bayes.

I generated a 500-row synthetic table — 125 rows each of metal can, plastic bottle, wooden block and rubber ball — with five features: distance, size, colour intensity, weight and sensor voltage. Each class has its own ranges for size, colour, weight and voltage. Distance is shared (15–90). The data is in data/synthetic_robotic_dataset.csv and can be rebuilt with the same seed from src/generate_dataset.py.

I scaled the features, encoded the labels, and held out 100 stratified rows for testing. Default KNN (k=3) scores 99% on that test set (one wooden block is called a rubber ball). Gaussian Naive Bayes scores 100%. After a small grid search both models reach 100% on the same 100 rows. Those figures are high because the classes were generated with fairly distinct weight and voltage ranges — this is a clean synthetic table, not a claim about real sensors.

The write-up I submitted is in report/report.pdf. It recorded 31% for both models. That run had trained on a first version of the table where labels were assigned at random, so the features had no relationship with the class. The scripts and notebook in this repo train on the class-conditional table described above.

Course: CS-4090 – Machine Learning for Robotics · Spring 2026 · FAST-NUCES · Assignment 4 · Roll: 22I-2327

---

What's Inside

Item	Path	Notes
Assignment brief	assignment/assignment-04.pdf	The original Assignment 4 statement
Submitted report	report/report.pdf, report/report.docx	The write-up I turned in (the 31% run)
Earlier Colab printout	report/notebook-colab-export.pdf	PDF export of an earlier notebook run
Notebook	notebooks/object_classification.ipynb	Full pipeline, with outputs
Dataset generator	src/generate_dataset.py	500 rows, seed 42
Training and evaluation	src/train.py	Preprocess, KNN, GaussianNB, grid search
Predict a new reading	src/predict.py	Reuses the saved scaler and class names
Dataset	data/synthetic_robotic_dataset.csv	The 500-row table
Screenshots	screenshots/	The captures used below

---

Dataset

500 rows, 125 per class, no missing values. Seed 42.

Class	Distance	Size	Colour intensity	Weight	Sensor voltage
metal_can	15–90	8–20	120–220	250–500	3.0–5.0
plastic_bottle	15–90	18–40	60–180	30–150	0.5–2.0
wooden_block	15–90	10–30	40–130	150–320	1.5–3.2
rubber_ball	15–90	6–18	150–255	80–220	1.0–2.8

Feature means on the generated table:

Class	Distance	Size	Colour	Weight	Voltage
metal_can	51.56	14.20	169.66	380.92	4.05
plastic_bottle	49.95	29.04	123.47	88.77	1.25
rubber_ball	52.29	12.04	203.02	151.99	1.86
wooden_block	51.94	20.51	84.52	237.70	2.35

Weight and voltage are the strongest separators. Wooden block and rubber ball overlap more than the others, which is where the one default-KNN mistake happened.

---

Models and Results

80/20 split, stratified, seed 42: 400 training rows and 100 test rows, 25 per class. Features are scaled with StandardScaler fitted on the training set only. Class names follow LabelEncoder order: metal_can, plastic_bottle, rubber_ball, wooden_block.

Default models

Model	Test accuracy	Notes
KNN (k=3)	99%	One wooden block predicted as rubber_ball
GaussianNB	100%	All 100 test rows correct

KNN (k=3) per class on the test set:

Class	Precision	Recall	F1
metal_can	1.0000	1.0000	1.0000
plastic_bottle	1.0000	1.0000	1.0000
rubber_ball	0.9615	1.0000	0.9804
wooden_block	1.0000	0.9600	0.9796

After grid search

Model	Best parameters	5-fold CV accuracy	Test accuracy
KNN	metric=manhattan, k=11, weights=distance	99.25%	100%
GaussianNB	var_smoothing=1e-09	99.75%	100%

On this test set the two tuned models tie at 100%. GaussianNB was already perfect with its defaults; the search kept the default smoothing and the 5-fold score is 99.75%. I would pick GaussianNB if I had to choose one here — it is already at 100% without extra knobs, and it is cheaper to evaluate than a 11-neighbour KNN.

src/predict.py applies the scaler that was fitted at training time. Fitting a new scaler on a single reading would collapse every feature to zero, so that is not used. A few checks:

python src/predict.py --distance 40 --size 12 --color 180 --weight 400 --voltage 4.2 --model knn
# metal_can

python src/predict.py --distance 40 --size 30 --color 100 --weight 80 --voltage 1.0 --model knn
# plastic_bottle

python src/predict.py --distance 50 --size 30 --color 120 --weight 300 --voltage 1.5 --model knn
# wooden_block

The fitted KNN, GaussianNB, scaler and label encoder are written together to models/object_classifiers.joblib when you run src/train.py. That file is left out of git on purpose — it is rebuilt in a few seconds.

---

Folder Structure

Object_Classification/
├── assignment/
│   └── assignment-04.pdf
├── report/
│   ├── report.pdf
│   ├── report.docx
│   └── notebook-colab-export.pdf
├── notebooks/
│   └── object_classification.ipynb
├── src/
│   ├── generate_dataset.py
│   ├── train.py
│   └── predict.py
├── data/
│   └── synthetic_robotic_dataset.csv
├── models/                         # created by src/train.py
├── screenshots/
├── assets/
│   ├── css/
│   │   └── style.css
│   └── js/
│       └── main.js
├── index.html
├── README.md
├── requirements.txt
├── LICENSE
├── .gitignore
├── .gitattributes
└── .nojekyll

---

How to Run

Python 3.10 or later. From the repository root:

python -m venv .venv

On Windows: .venv\Scripts\activate
On Linux or macOS: source .venv/bin/activate

pip install -r requirements.txt
python src/generate_dataset.py
python src/train.py
python src/predict.py --distance 40 --size 12 --color 180 --weight 400 --voltage 4.2 --model knn

generate_dataset.py is optional if you just want to reuse the CSV already in data/. train.py will also generate the table if the CSV is missing.

The notebook is notebooks/object_classification.ipynb. Open it from the repository root, or from notebooks/ — it looks for src/ and data/ in both places.

To preview the website locally, open index.html or serve the folder:

python -m http.server 8080

then visit http://localhost:8080.

---

Screenshots

1. Repository Folder Structure

2. Assignment Brief

3. Dataset Sample and Class Counts

4. Feature Means by Class

5. Weight and Voltage by Class

6. Feature Ranges by Class

7. KNN Default Classification Report

8. KNN Default Confusion Matrix

9. GaussianNB Classification Report

10. GaussianNB Confusion Matrix

11. Grid Search and Tuned Comparison

12. Classifying a New Sensor Reading

13. Notebook Walkthrough

14. Website Home (Dark Theme)

15. Website Home (Light Theme)

16. Website on a Phone

17. Results Section on the Website

18. How to Run on the Website

---

License

This project is released under the MIT License. The assignment brief in assignment/ and the report in report/ remain course material and are included as I received or submitted them.

---

Author

Mohammad Rohaan
FAST-NUCES · 22I-2327
github.com/rohaan2802