# 2025 IEEE ICIP VIP CUP — Data

来源：https://www.kaggle.com/competitions/2025-ieee-icip-vip-cup/data

采集时间：2026-10-25T08:35:04Z

# Dataset Description

This section describes exactly the datasets provided for the competition, based solely on the official document. It covers dataset splits, modalities, counts, and organization as specified in Section 2.2 and Section 2.5 of the Official Document .

---

## 1. Overview

The competition provides two primary datasets, as detailed in Section 2.2:

1. **Drone Detection and Tracking Dataset**  
   - **Type(s)**: IR/Thermal, RGB/Visual  
   - **Modality**: Images and Videos  
   - **Number of classes**: 2 (Drone and Bird)  
   - **Train set**: 45,000 IR–RGB image pairs  
   - **Validation set**: 6,500 IR and RGB images + 40 video sequences  
   - **Test set**: 13,000 IR and RGB images + 25 video sequences  
   - **Frame rate of videos**: 30 fps  
   - **Resolution of each video**: 320 × 256 pixels  
   - **Image size**: 320 × 256 pixels  
   - **Duration of each video**: 10 s  
   - **Environmental factors considered**: Hilly regions, thick forest cover, cloudy sky, fog and mist  
   - **Scenarios**: Single class in a frame, multiple classes in a frame, objects far off from the FoV of the source camera, swarm of birds and drones, drones captured under extreme environmental conditions as stated above  
   - **Distortions considered**: Speckle noise, salt and pepper noise, Gaussian blur, uneven illumination, motion blur, camera instability and AWGN  
   - **Number of videos with distortions**: 30  
   - **Intensity of distortions (on a scale of 5)**:
     - Speckle noise – 2  
     - Salt and Pepper noise – 2  
     - Gaussian blur – 4  
     - Uneven illumination – 4  
     - Motion blur – 4  
     - Camera instability – 5  
     - AWGN – 3  
   :contentReference[oaicite:5]{index=5}

2. **Payload Identification Dataset**  
   - **Type(s)**: IR/Thermal, RGB/Visual  
   - **Modality**: Images  
   - **Number of classes**: 2 (Harmful payload and Normal payload)  
   - **Train set**: 25,000 IR–RGB image pairs  
   - **Validation set**: 4,000 IR and RGB images  
   - **Test set**: 8,000 IR and RGB images  
   - **Image size**: 320 × 256 pixels  
   - **Environmental factors considered**: Hilly regions, thick forest cover, cloudy sky, fog and mist  
   - **Scenarios**: Single class in a frame, multiple classes in a frame, objects far off from the FoV of the source camera, swarm of birds and drones, drones captured under extreme environmental conditions as stated above  
   - **Distortions considered**: Speckle noise, salt and pepper noise, Gaussian blur, uneven illumination, motion blur, camera instability and AWGN  
   - **Intensity of distortions (on a scale of 5)**:
     - Speckle noise – 2  
     - Salt and Pepper noise – 2  
     - Gaussian blur – 4  
     - Uneven illumination – 4  
     - Motion blur – 4  
     - Camera instability – 5  
     - AWGN – 3  
   :contentReference[oaicite:6]{index=6}

*(All values above are taken verbatim from Section 2.2 “Dataset Description” of the Official Document.)*

---

## 2. Data Organization

As specified in Section 2.5 of the Official Document, the directory layout and file organization are as follows:

- **Main Directory**: `Data/`  
  - **RGB/**  
    - **Train/**: Folder containing `Images/` and `Labels/`  
    - **Validation/**: Folder containing `Images/` and `Labels/`  
    - **Test/**: Folder containing `Images/` and `Labels/`  
  - **IR/**  
    - **Train/**: Folder containing `Images/` and `Labels/`  
    - **Validation/**: Folder containing `Images/` and `Labels/`  
    - **Test/**: Folder containing `Images/` and `Labels/`  
  - **Payload/**  
    - **Train/**: Folder containing `Images/` and `Labels/`  
    - **Validation/**: Folder containing `Images/` and `Labels/`  
    - **Test/**: Folder containing `Images/` and `Labels/`  
- **Additional files in the root**:
  - `Dataset_Statistics.txt` – File describes dataset statistics  
  - `Challenge_Description.txt` – File describes the challenge  
:contentReference[oaicite:7]{index=7}

> **Note**: The actual filenames within `Images/` and `Labels/` folders (e.g., exact image file naming conventions, label file formats) are provided in these subfolders; this section reflects only the high-level folder structure as given.

---

## 3. How to Present in Kaggle “Data” Tab

Copy the above sections under your Kaggle competition Data page or README.md:

1. **Header**: `# Dataset Description`
2. **Subsections**:  
   - `## 1. Overview` containing the two dataset descriptions (Drone Detection and Tracking Dataset; Payload Identification Dataset) with exact parameters.  
   - `## 2. Data Organization` with the directory layout as above.

No other details or placeholder values are included—only what the official document specifies.

---

## Example Kaggle Markdown Block

```markdown
# Dataset Description

This section describes exactly the datasets provided for the competition, based solely on the official document.

---

## 1. Overview

The competition provides two primary datasets:

1. **Drone Detection and Tracking Dataset**  
   - **Type(s)**: IR/Thermal, RGB/Visual  
   - **Modality**: Images and Videos  
   - **Number of classes**: 2 (Drone and Bird)  
   - **Train set**: 45,000 IR–RGB image pairs  
   - **Validation set**: 6,500 IR and RGB images + 40 video sequences  
   - **Test set**: 13,000 IR and RGB images + 25 video sequences  
   - **Frame rate of videos**: 30 fps  
   - **Resolution of each video**: 320 × 256 pixels  
   - **Image size**: 320 × 256 pixels  
   - **Duration of each video**: 10 s  
   - **Environmental factors considered**: Hilly regions, thick forest cover, cloudy sky, fog and mist  
   - **Scenarios**: Single class in a frame, multiple classes in a frame, objects far off from the FoV of the source camera, swarm of birds and drones, drones captured under extreme environmental conditions as stated above  
   - **Distortions considered**: Speckle noise, salt and pepper noise, Gaussian blur, uneven illumination, motion blur, camera instability and AWGN  
   - **Number of videos with distortions**: 30  
   - **Intensity of distortions (on a scale of 5)**:
     - Speckle noise – 2  
     - Salt and Pepper noise – 2  
     - Gaussian blur – 4  
     - Uneven illumination – 4  
     - Motion blur – 4  
     - Camera instability – 5  
     - AWGN – 3  

2. **Payload Identification Dataset**  
   - **Type(s)**: IR/Thermal, RGB/Visual  
   - **Modality**: Images  
   - **Number of classes**: 2 (Harmful payload and Normal payload)  
   - **Train set**: 25,000 IR–RGB image pairs  
   - **Validation set**: 4,000 IR and RGB images  
   - **Test set**: 8,000 IR and RGB images  
   - **Image size**: 320 × 256 pixels  
   - **Environmental factors considered**: Hilly regions, thick forest cover, cloudy sky, fog and mist  
   - **Scenarios**: Single class in a frame, multiple classes in a frame, objects far off from the FoV of the source camera, swarm of birds and drones, drones captured under extreme environmental conditions as stated above  
   - **Distortions considered**: Speckle noise, salt and pepper noise, Gaussian blur, uneven illumination, motion blur, camera instability and AWGN  
   - **Intensity of distortions (on a scale of 5)**:
     - Speckle noise – 2  
     - Salt and Pepper noise – 2  
     - Gaussian blur – 4  
     - Uneven illumination – 4  
     - Motion blur – 4  
     - Camera instability – 5  
     - AWGN – 3  

*(All values above are taken verbatim from Section 2.2 of the Official Document.)*

---

## 2. Data Organization

As specified in Section 2.5 of the Official Document:

- **Main Directory**: `Data/`  
  - **RGB/**  
    - **Train/**: Folder containing `Images/` and `Labels/`  
    - **Validation/**: Folder containing `Images/` and `Labels/`  
    - **Test/**: Folder containing `Images/` and `Labels/`  
  - **IR/**  
    - **Train/**: Folder containing `Images/` and `Labels/`  
    - **Validation/**: Folder containing `Images/` and `Labels/`  
    - **Test/**: Folder containing `Images/` and `Labels/`  
  - **Payload/**  
    - **Train/**: Folder containing `Images/` and `Labels/`  
    - **Validation/**: Folder containing `Images/` and `Labels/`  
    - **Test/**: Folder containing `Images/` and `Labels/`  
- **Additional files** in root:
  - `Dataset_Statistics.txt` – File describes dataset statistics  
  - `Challenge_Description.txt` – File describes the challenge  

*(Folder and file names are exactly as provided in the Official Document.)*