eidolon08/yolo_seg_labeled_data_for_small_toycar

---
license: cc-by-4.0
task_categories:
- object-detection
tags:
- yolo
- autonomous-driving
- parking
- small-car
- toycar
- roboflow
size_categories:
- 10K<n<100K
---

# YOLO Segmentation Labeled Data for Small Toy Car

This is an autonomous driving YOLO object detection dataset for small RC/toy cars.
It contains bounding box annotations on images captured in parking and track driving environments.

## Dataset Structure

The dataset consists of a total of **4 subsets** and approximately **75,000 images**.
Each subset includes `train / valid / test` splits and is provided as a `tar.gz` archive.

| Subset | Description | Number of Images | Number of Classes |
|--------|-------------|------------------|-------------------|
| `parking_front` | Front parking environment | 10,719 | 3 |
| `parking_rear` | Rear parking environment | 5,639 | 3 |
| `track` | Track driving environment | 20,135 | 4 |
| `track_crosswalk` | Track crosswalk environment | 38,692 | 1 |

---

## Class Information

### parking_front
| ID | Class Name | Description |
|----|------------|-------------|
| 0 | `out_line` | Parking zone outline |
| 1 | `parking_lot` | Parking lot area |
| 2 | `parking_space` | Individual parking space |

### parking_rear
| ID | Class Name | Description |
|----|------------|-------------|
| 0 | `end_line` | Parking end line |
| 1 | `parking_lot` | Parking lot area |
| 2 | `parking_space` | Individual parking space |

### track
| ID | Class Name | Description |
|----|------------|-------------|
| 0 | `car` | Car |
| 1 | `lane1` | Lane 1 |
| 2 | `lane2` | Lane 2 |
| 3 | `traffic_light` | Traffic light |

### track_crosswalk
| ID | Class Name | Description |
|----|------------|-------------|
| 0 | `crosswalk` | Crosswalk |

---

## How to Use

### 1. Download and Extract

```bash
# Install huggingface_hub
pip install huggingface_hub

# Download the desired subset (e.g., track)
from huggingface_hub import hf_hub_download

path = hf_hub_download(
repo_id="eidolon08/yolo_seg_labeled_data_for_small_toycar",
filename="track.tar.gz",
repo_type="dataset"
)

Or via CLI:

Bash
huggingface-cli download eidolon08/yolo_seg_labeled_data_for_small_toycar \
track.tar.gz --repo-type dataset --local-dir ./data
Extract the archive:

Bash
tar -xzf track.tar.gz
2. Directory Structure
The structure of each subset after extraction is as follows:

{subset}/
├── data.yaml
├── train/
│ ├── images/
│ │ ├── chunk_0000/ # Image files (in chunks of 1,000)
│ │ ├── chunk_0001/
│ │ └── ...
│ └── labels/
│ ├── chunk_0000/ # YOLO format annotations (.txt)
│ ├── chunk_0001/
│ └── ...
├── valid/
│ ├── images/
│ └── labels/
└── test/
├── images/
└── labels/
3. YOLO Training
Modify the paths in data.yaml to fit your local environment and start training:

YAML
train: ./track/train/images
val: ./track/valid/images
test: ./track/test/images

nc: 4
names: ['car', 'lane1', 'lane2', 'traffic_light']
Bash
yolo detect train data=data.yaml model=yolov8n.pt epochs=100
Annotation Format
The dataset uses the YOLO format (.txt). Each line represents one object:

<class_id> <x_center> <y_center> <width> <height>
All values are normalized (0~1) relative to the image dimensions.