Building Wildfire Smoke Classifier via Transfer Learning
Install Docker:
https://docs.docker.com/v17.12/docker-for-mac/install/#install-and-run-docker-for-mac
We use the scripts provided by the following excellent Google’s TensorFlow For Poets codelab. https://codelabs.developers.google.com/codelabs/tensorflow-for-poets/#0
Increase the memory available to Docker Engine
- Navigate to Docker Preferences
- Select Advanced
- Increase Memory to 4.0 GiB
Clone the repository (https://github.com/aiformankind/wildfire-smoke-classifier-camera.git):
git clone https://github.com/aiformankind/wildfire-smoke-classifier-camera.git
Go to the repository directory that you just clone:
cd wildfire-smoke-classifier-camera
Build the Tensorflow docker (this job will pull the latest tensorflow images and set up the environment) :
docker build -t aiformankind/wildfire-smoke-camera:0.0.1 .
Start the Tensorflow container (this job will spin up the seetheworld container):
docker run -it -p 8888:8888 -p 6006:6006 --name=wildfire-smoke-camera aiformankind/wildfire-smoke-camera:0.0.1
Augment data:
python augment/augment_images.py --image_dir=data/usa/wildfire_smoke --num_samples=3000
Retrain model:
python -m train.retrain --bottleneck_dir=train_output/bottlenecks --how_many_training_steps=500 --model_dir=train_output/models/ --summaries_dir=train_output/training_summaries/"${ARCHITECTURE}" --output_graph=train_output/retrained_graph.pb --output_labels=train_output/retrained_labels.txt --architecture="${ARCHITECTURE}" --image_dir=augment-data/usa/wildfire_smoke
Predict label:
python -m train.label_image --graph=train_output/retrained_graph.pb --image=validation/usa/wildfire_smoke/00000001.jpg
- Problem: The Augment data step fails with error 'Killed'. Resolution: Increase memory allocated to Docker Engine and rerun.