Tensorflow image recognition

Object recognition with use of Tensorflow. Based on the retrain example: https://github.com/tensorflow/tensorflow/blob/master/tensorflow/examples/image_retraining/retrain.py

RQT tools related to this package can be found in package image_recognition_tensorflow_rqt

Installation

See https://github.com/tue-robotics/image_recognition for installation instructions. Also make sure that you have a working installation on Tensorflow. If you are running Ubuntu x64, Tensorflow can be installed with use of the following commands:

sudo apt-get install python-pip python-dev
sudo pip install tensorflow # If you want gpu support, use pip install tensorflow-gpu

Quick How-to

1. Annotate images with use of annotation tool provided in https://github.com/tue-robotics/image_recognition/tree/master/image_recognition_rqt
2. Retrain the neural network: https://github.com/tue-robotics/image_recognition/tree/master/image_recognition_tensorflow_rqt
3. Start the ROS node with the net as parameter:

    rosrun image_recognition_tensorflow object_recognition_node _graph_path:=[path_to_graph.pb] _labels_path:=[path_to_labels.txt]
    

1. Test the classifier with use the test tool in https://github.com/tue-robotics/image_recognition/tree/master/image_recognition_rqt

ROS Node (object_recognition_node)

rosrun image_recognition_tensorflow object_recognition_node _graph_path:=[path_to_graph.pb] _labels_path:=[path_to_labels.txt]

Scripts

Get object recognition (get_object_recognition)

Get the classification result of an input image:

rosrun image_recognition_tensorflow get_object_recognition [path_to_graph.pb] [path_to_labels.txt] [path_to_image]

Evaluate classifier (evaluate_classifier)

Evaluate the classifier based on a structured images folder. The script assumes that the images in this directory are separated in different directories with the label name as directory name.

rosrun image_recognition_tensorflow evaluate_classifier [path_to_graph.pb] [path_to_labels.txt] [path_to_image_dir]

Retrain the neural network (retrain)

Train the neural network based on a set of images. The script assumes that the images in this directory are separated in different directories with the label name as directory name.

rosrun image_recognition_tensorflow retrain [image_folder] [model_folder_inceptionv3] [output_dir]

\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^ Changelog for package image_recognition_tensorflow \^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^\^

0.0.5 (2019-06-08)

• Added test for image_recognition_tensorflow
• Added catkin lint to all packages
• refactor package xml to 2.0
• rename ROS pkgs with image_recognition_prefix
• Contributors: Matthijs van der Burgh, Rein Appeldoorn

0.0.4 (2017-02-07)

0.0.3 (2017-02-07)

• fix(): Add targets install for scripts
• Add numpy run dependency for image_recognition_tensorflow
• Contributors: Matthijs van der Burgh, Rein Appeldoorn

0.0.2 (2017-01-17)

0.0.1 (2017-01-17)

• Add try except if path already exists
• Update retrain to the latest version of tensorflow
• Added image recognition util to support labeled and raw writing of image in predefined folder structure
• Update README.md
• Update README.md
• Documentation
• Working annotation gui and test gui for both image_recognition_tensorflow object recognition and image_recognition_openface face recognition
• Working with updated interface
• Update README.md
• os.path.expanduser for paths
• Some minor reference and naming fixes
• Merged tensorflow_object_recognition into image_recognition repo, renamed to image_recognition_tensorflow
• Contributors: Ramon Wijnands, Rein Appeldoorn