SOFTWARE & SYSTEMS
FIRE & SMOKE DETECTION CNN
A Keras image classifier that detects fire and smoke in images, using transfer learning on a VGG16 base.
completed
- Year
- 2019
- Framework
- Keras
POSITIVE CLASS SAMPLE, FIRE
VGG16 TRANSFER-LEARNING IMAGE CLASSIFIER
01 / The build
Behind the project.
A fire and smoke image classifier built in Keras: a frozen ImageNet-pretrained VGG16 base with a dense head, trained on a fire/smoke dataset in three classes (fire, neutral, smoke) at 150x150 input with data augmentation, early stopping and checkpointing. The repo includes training, continued-training, augmentation and test scripts.
02 / In the workshop
Project media.
3 photos & media · Tap to expand
Technical details
Under the hood.
- Framework
- Keras
- Base
- VGG16 (ImageNet, Frozen) + Dense Head
- Classes
- Fire / Neutral / Smoke
- Input
- 150x150
- Training
- Augmentation, Early Stopping, Checkpointing
Notes & outcomes
What came out of it.
- Transfer learning on VGG16.
- Three-class fire/smoke/neutral classifier.
Keep exploring