3DLAS

# 3DLAS Dataset

## Overview
The 3DLAS dataset is a novel collection of point cloud data captured by LAS-LIDAR (Linear Array Sensing LiDAR), an innovative industrial linear LiDAR system for real-time 3D mapping and object recognition from shape. This dataset is designed to support research in 3D object detection and classification, particularly for sparse and uniquely structured point clouds generated by linear array sensors. Here = we provide  

## Dataset Characteristics
- **Sensor Type**: LAS-LIDAR (Linear Array Sensing LiDAR)
- **Classes**: 11 distinct classes
 - balloon, table, chair, bicycle, bicycle with riders, fire extinguisher, protective helmet, person, coat rack, wall, and whiteboard on wheels
- **Point Cloud Density**: 2048 points per sample
- **File Format**: CSV
- **Data Type**: Raw, non-augmented data

## Directory Structure
```
3DLAS/
│
├── balloon/
├── table/
├── chair/
├── bicycle/
├── bicycle_with_riders/
├── fire_extinguisher/
├── protective_helmet/
├── person/
├── coat_rack/
├── wall/
└── whiteboard_on_wheels/
```
Each directory contains multiple CSV files representing individual samples for that class.

## Data Format
- Point Cloud Files: CSV format with three columns
 - [timestamp, sensor_index, distance]
- Each file represents a single object instance with 2048 points

## Usage
To use this dataset:
1. Download the repository
2. Load the CSV files using your preferred data processing library (e.g., pandas, numpy)
3. Process the point cloud data as needed for your 3D object detection or classification tasks

Example Python code to load a sample:

import pandas as pd

# Load a single sample
sample = pd.read_csv('3DLAS/chair/file1.csv', 
                    names=['timestamp', 'sensor_index', 'distance'])

# Convert to numpy array if needed
sample_array = sample.values

## Data Augmentation
The provided data is non-augmented. For data augmentation, you can use the data_augmentation.py file.