Remote-Sensing-Image-Recognition

Application of Convolutional Neural Network in Remote Sensing Image Recognition

  1. Convert the Texas Image into PNG format and split it into small tiles

2.Start Docker with local files available % docker run -it -v $HOME/tf_files:/tf_files gcr.io/tensorflow/tensorflow:latest-devel # my file: /tf_filesclassification/evergreen forest, shrub_scrub, cultivated_crops

3.Retrieving the training code cd /tensorflow git pull

4.Start your image retraining python tensorflow/examples/image_retraining/retrain.py \ # retrain contains the code used to train image classification --bottleneck_dir=/tf_files/bottlenecks
--how_many_training_steps 500
--model_dir=/tf_files/inception
--output_graph=/tf_files/retrained_graph.pb
--output_labels=/tf_files/retrained_labels.txt
--image_dir /tf_files/classification

5.use the label_image.py to load my graph file and predicts with it. I put label_image.py in the folder of ty_files.

python /tf_files/label_image.py /tf_files/testpicture/*picture name