Vision project

Set up of TF object detection module.

Please remember to put this in bin/activate after following the tensorflow model detection api’s obj:

https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/installation.md#add-libraries-to-pythonpath

# INstallation instruction quick summary
# For CPU
pip install tensorflow
# For GPU
pip install tensorflow-gpu

The remaining libraries can be installed on Ubuntu 16.04 using via apt-get:

sudo apt-get install protobuf-compiler python-pil python-lxml python-tk
pip install --user Cython
pip install --user contextlib2
pip install --user jupyter
pip install --user matplotlib

Alternatively, users can install dependencies using pip:

pip install --user Cython
pip install --user contextlib2
pip install --user pillow
pip install --user lxml
pip install --user jupyter
pip install --user matplotlib

# FOr coco metrics
git clone https://github.com/cocodataset/cocoapi.git
cd cocoapi/PythonAPI
make
cp -r pycocotools <path_to_tensorflow>/models/research/

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


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

Experiments performed

The data was prepared (made into tf-record shards of size 10) according to the guide.

The following files were edited to allow for setting up the pipeline and the hyper-parameters for faster RCNN as well as what was to be done during training.

config/faster_rcnn_resnet101_sdc.config – has the configuration. config/sdc_label_map.pbtxt – Has the class to label mapping.

utils/visualize allows one to visualize tf-record shards. Sharding was done to fit the training data onto the GPU.

the run_sdc_train.sh script is used for running a tensorflow object detection. This will have to be from the tensorflow’s model/research folder.

GCP will also have to be set up according to the installation guide.

Task 2

Task 2 Instructions: Run the file Task2FinalCode.py with bbox_data in the code pointing to the correct location of the bounding box pkl file. The code generates outputTask2.csv file in the current working directory.