Grab and Go Dataset

Source Code for Automate Generate tfrecord from Kaggle Grab and Go Dataset

Installation

set up your python environment and use package manager pip to install requirement depedency

pip install -r requirements.txt

then install tensorflow object detection api using our setupapi

git clone https://github.com/tensorflow/models
cd models/research/
protoc object_detection/protos/*.proto --python_out=.
cp object_detection/packages/tf2/setup.py .
python -m pip install .

Download Kaggle Dataset

Download the kaggle dataset from our collected retail data in this link

Extract the data so the project repository structure look like this

.
├── data/                   # CSV file location
├── annotations/            # Extracted Dataset
    └── train/
    └── test/
├── images/                 # Extracted Dataset
    └── train/
    └── test/                 
├── generate_tfrecord.py                     
├── requirements.txt                    
├── detection_label_map.pbtxt                   
├── LICENSE
└── README.md

Convert XML Annotations Dataset

to generate the tfrecords formated input we need to convert XML annotations data to CSV data by running

python xml_to_csv.py

Generate TFRECORDS

We use tfrecords as input for our model training, to generate it we run generate_tfrecord.py by running

python generate_tfrecord.py --csv_input=data/train_labels.csv  --output_path=train.record --image_dir=images/train/
python generate_tfrecord.py --csv_input=data/test_labels.csv  --output_path=test.record --image_dir=images/test/

Model Training

Access the Colab Notebook in this link to train Grab and Go Retail Detection Model