Aerial-Detection-Faster-RCNN

Human detection for Search and Rescue operation in UAV’s using Faster RCNN Model

Installations

Dependencies

Tensorflow Object Detection API depends on the following libraries:

• Protobuf 2.6
• Python-tk
• Pillow 1.0
• lxml
• tf Slim (which is included in the "tensorflow/models/research/" checkout)
• Jupyter notebook
• Keras
• Matplotlib
• Tensorflow

Tensorflow Models

https://github.com/tensorflow/models

# For CPU
pip install tensorflow
# For GPU
pip install tensorflow-gpu

Datasets

Dataset used for Aerial [https://www.kaggle.com/misteryprsm/aerial-dataset] object detection

Protobuf Compilation

The Tensorflow Object Detection API uses Protobufs to configure model and training parameters. Before the framework can be used, the Protobuf libraries must be compiled. This should be done by running the following command from the tensorflow/models/research/ directory:

# From tensorflow/models/research/
protoc object_detection/protos/*.proto --python_out=.

Add Libraries to PYTHONPATH

When running locally, the tensorflow/models/research/ and slim directories should be appended to PYTHONPATH. This can be done by running the following from tensorflow/models/research/:

# From tensorflow/models/research/
export PYTHONPATH=$PYTHONPATH:`pwd`:`pwd`/slim

Testing the Installation

You can test that you have correctly installed the Tensorflow Object Detection
API by running the following command:

python object_detection/builders/model_builder_test.py

Configuring the Object Detection Training Pipeline

The Tensorflow Object Detection API uses protobuf files to configure the training and evaluation process. The schema for the training pipeline can be found in object_detection/protos/pipeline.proto.

A skeleton configuration file is shown below:

model {
(... Add model config here...)
}

train_config : {
(... Add train_config here...)
}

train_input_reader: {
(... Add train_input configuration here...)
}

eval_config: {
}

eval_input_reader: {
(... Add eval_input configuration here...)
}

Defining Inputs

The Tensorflow Object Detection API accepts inputs in the TFRecord file format. Users must specify the locations of both the training and evaluation files. Additionally, label map should be specified , which define the mapping between a class id and class name. The label map should be identical between training and evaluation datasets.

An example input configuration looks as follows:

tf_record_input_reader {
  input_path: "/usr/home/username/training_data/train.record"
}
label_map_path: "/usr/home/username/data/label_map.pbtxt"

Preparing Inputs

Tensorflow Object Detection API reads data using the TFRecord file format.

Generating the TFRecord files.

Images are labelled using lblImg [https://github.com/tzutalin/labelImg] which is in xml format.
XML files are then converted to csv format using sml_to_csv.py [https://github.com/datitran/raccoon_dataset/blob/master/xml_to_csv.py]

Now the folder structure for object detection will look as follows

Object-Detection
-training_data/
--test_labels.csv
--train_labels.csv
--test_images/
---testingimages.jpg
--...yourimages.jpg
--training_images/
---testingimages.jpg
--...yourimages.jpg

To convert these into TFRecords, run the following commands:

python generate_tfrecord.py --csv_input=training_data/train_labels.csv --output_path=training_data/train.record

python generate_tfrecord.py --csv_input=training_data/test_labels.csv --output_path=training_data/test.record

TensorFlow model

model used for object detection is coco trained faster_rcnn_inception_v2 which can be downloaded from [https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md] with the configuration file ssd_mobilenet_v1_pets.config from [https://github.com/tensorflow/models/tree/master/research/object_detection/samples/configs]

start the training with

python train.py --logtostderr --train_dir=training/ --pipeline_config_path=training/faster_rcnn_inception_v2_coco.config

From models/object_detection, via terminal, start TensorBoard with:

tensorboard --logdir='training'