FireDetection is an artificial intelligence project for real-time fire detection and custom images.

This is the first release of the FireDetection. It contains dataset of 42 test images for training 100 images.

To use your fire dataset in Tensorflow Object Detection API, you must convert it into the TFRecord file format. Dataset into a images document.

Each dataset is required to have a label map associated with it. This label map defines a mapping from string class names to integer class Ids. The label map should be a StringIntLabelMap text protobuf. Sample label maps can be found in object_detection/training/labelmap. Label maps should always start from id 1.

If you want train this model :

python train.py --logtostderr --train_dir=training/ --pipeline_config_path=training/faster_rcnn_inception_v2_pets.config

After convert model checkpoint to pb file :

python export_inference_graph.py --input_type image_tensor --pipeline_config_path training/faster_rcnn_inception_v2_pets.config --trained_checkpoint_prefix training/model.ckpt-6516 --output_directory inference_graph

pip install requirements.txt 

• If you have an Nvidia GPU, then you can install tensorflow-gpu package. It will make things run a lot faster. Depending on the hardware configuration of your system, the execution time will vary. On CPU, training will be slow.If you want realtime fire detection model effect? Please use only tensorflow_gpu GPU .

• Minimal:

• 16gb RAM

• Core i7

• Nvidia

python Object_detection_image.py

collect data from Google: Fire and Smoke

Train Data : Train data contains the 200 images of each fire,smoke total their are 400 images in the training dataset Test Data : Test data contains 50 images of each fire and smoke total their are 100 images in the test dataset

Below is the complete implementation:

 loss = 0.018

 > cd Keras_model
 > Directory of G:\FireDetection\object_detection\Keras_model
 > jupyter notebook firedetection.ipynb

If you are facing any difficulty, feel free to create a new issue or reach out on email [email protected]