Fire-Flame-Dataset

---

An image dataset for training fire and frame detection AI

---

Fire-Flame-Dataset is a dataset collected in order to train machine learning model to recognize Fire, smoke, and neutral(images without fire or smoke).This a dataset containing about 3000 images and 3 classes which include:

• Fire
• Smoke
• neutral

There are 1000 images in each category and 900 for train and 100 for testing

Download, Training and Prediction

---

The Fire-Flame-Dataset is provided for download in the release section of this repository. You can download the dataset via this link Fire-Flame-Dataset.

The implementation code in which the model was train with has been provide in this repository. The model was trained with train with resnet50 and a accuracy of 85% on the test data was achieved. The python codebase is contained in fire_flame.ipynb.

Some of the prediction results are shown below:

('Image of:', 'Class: Fire', 'Confidence score: 1.0')

('Image of:', 'Class: Fire', 'Confidence score: 0.990234375')

('Image of:', 'Class: Neutral', 'Confidence score: 0.99365234375')

('Image of:', 'Class: Neutral', 'Confidence score: 1.0')

('Image of:', 'Class: Smoke', 'Confidence score: 0.4462890625')

('Image of:', 'Class: Smoke', 'Confidence score: 0.9970703125')

Reqirements

---

• Python 3
• Pytorch
• Numpy
• Matplotlib
• TorchFussion

References

---

• Kaiming H. et al, Deep Residual Learning for Image Recognition