The KIIT Military Target Archive (KIIT-MiTA) dataset is a high-resolution drone-captured image dataset designed for military object detection and recognition. It provides carefully annotated images in the YOLO format, enabling researchers and developers to train models for real-time surveillance, object detection, and tracking in military applications.
- Total Images: 1,700
- Annotation Format: YOLO
- Total Annotations: 4,100+
- Classes: 7 distinct military objects
- Resolution: High-resolution drone imagery
- Sources: Publicly available data, YouTube frames, and self-collected images
The dataset includes 7 military object categories:
- Artillery
- Missile
- Radar
- M. Rocket Launcher
- Soldier
- Tank
- Vehicle
Each image is labeled with bounding boxes and class annotations, ensuring precise object detection capabilities.
You can access the dataset from multiple sources:
- 📌 Kaggle: KIIT-MiTA Dataset
- 📌 Google Drive: Download Here
- 📌 Mega: Mega Download
- Collected from publicly available military datasets, YouTube video frames, and self-captured drone images.
- Covers various lighting conditions, weather scenarios, and terrains to enhance model generalization.
- Labeled using CVAT (Computer Vision Annotation Tool).
- 4,100+ object annotations were manually reviewed for accuracy.
- Each annotation includes bounding box coordinates and class labels, normalized for YOLO model compatibility.
- 📸 High-resolution drone imagery for precise military object detection
- 🏗 Manually annotated dataset optimized for deep learning models
- 🔄 Robust augmentation techniques applied for better generalization
- 🏆 Split into Training (80%), Validation (10%), and Testing (10%)
- 💡 Ideal for YOLO, Faster R-CNN, SSD, and other object detection models
- ✅ Educational & Research Use Only
- 🚫 Strictly Prohibited for Commercial Use
- 🔗 Must credit the authors if used in research/publications
If you use this dataset in your research, please cite our work and provide proper attribution.
This dataset was created by researchers from KIIT University:
| Name | GitHub Profile | LinkedIn Profile | |
|---|---|---|---|
| Sudip Chakrabarty | sudipchakrabarty6@gmail.com | Sudip-329 | |
| Sourov Roy Shuvo | sourovroyshuvo777@gmail.com | SourovRS | |
| Rajesh Chowdhury | rajesh99.bd@gmail.com | rajeshbd99 | |
| Sorup Chakraborty | sorupchakraborty001@gmail.com | sorupchakraborty |
👨🏫 Dr. Rajdeep Chatterjee
Associate Professor, School of Computer Engineering, KIIT
🔗 LinkedIn
This dataset is available under the Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) License.
🔗 More details: CC BY-NC 4.0 License
If you use this dataset, please cite our research paper:
@INPROCEEDINGS{10969335,
author={Chakrabarty, Sudip and Chatterjee, Rajdeep and Chakraborty, Sorup and Roy Shuvo, Sourov and Chowdhury, Rajesh},
booktitle={2025 3rd International Conference on Intelligent Systems, Advanced Computing and Communication (ISACC)},
title={Drones in Defense: Real-Time Vision-Based Military Target Surveillance and Tracking},
year={2025},
volume={},
number={},
pages={508-513},
keywords={Training;Target tracking;Accuracy;Military computing;Surveillance;Computational modeling;Radar tracking;Real-time systems;Drones;Videos;Drone;KIIT-MiTA;Military;Object Detection;YOLOv11;Tracking},
doi={10.1109/ISACC65211.2025.10969335}}
---
## 🛠 Recommended Usage
This dataset is **ideal for training deep learning-based object detection models** such as:
- **YOLOv4, YOLOv5, and YOLOv8**
- **Faster R-CNN**
- **SSD (Single Shot MultiBox Detector)**
- **EfficientDet**
- **Vision Transformers (ViTs)**
---
## 📬 Contact
For any questions or contributions, feel free to reach out via email or connect on LinkedIn.
📧 **Email:** rajesh99.bd@gmail.com
📧 **Email:** sudipchakrabarty6@gmail.com
🔗 **GitHub Repo:** [KIIT-MiTA](https://github.com/Sudip-329/KIIT-MiTA)
---
## ⭐ Acknowledgment
We appreciate the support of **KIIT University** and **Dr. Rajdeep Chatterjee** sir for facilitating this research and providing computational resources for dataset preparation.
---