swati6singh/ss20478_Drone_Detection

---
language:
- en
license: mit
task_categories:
- object-detection
---

# Drone Detection Frames

This dataset contains annotated video frames where a drone was detected using a fine-tuned YOLOv8s model.
Each row is a single frame with at least one drone detection, stored with its bounding box coordinates and confidence scores.

## Dataset Structure

| Column | Type | Description |
|---|---|---|
| `image` | `Image` | Annotated frame with bounding boxes drawn in green |
| `filename` | `string` | Original filename e.g. `drone_video_1_frame_00042.jpg` |
| `video_name` | `string` | Source video name |
| `frame_idx` | `int32` | Frame number within the original video |
| `frame_width` | `int32` | Frame width in pixels |
| `frame_height` | `int32` | Frame height in pixels |
| `num_detections` | `int32` | Number of drones detected in this frame |
| `box_x1` | `Sequence[int32]` | Left edge of each bounding box |
| `box_y1` | `Sequence[int32]` | Top edge of each bounding box |
| `box_x2` | `Sequence[int32]` | Right edge of each bounding box |
| `box_y2` | `Sequence[int32]` | Bottom edge of each bounding box |
| `box_conf` | `Sequence[float32]` | Confidence score per box |
| `conf_threshold` | `float32` | Confidence threshold used during inference |