Tiny Pascal VOC Instance Segmentation

The proposed challenge is a tiny Pascal VOC instance segmentation. Contains two parts:

1. Do segmentation mask for each instance
2. classify the instances into 20 classes

The giving tiny Pascal VOC dataset contains 1349 images for training and 100 images for testing. This project uses the Pytorch library and backbone pre-trained on ImageNet model to fix this challenge.

Hardware

• Intel(R) Core(TM) i5-9600K CPU @ 3.70GHz
• NVIDIA GeForce RTX 2080 Ti

Environment

• Microsoft win10
• Python 3.7.3
• Pytorch 1.7.0
• CUDA 10.2

Reproducing Submission

To reproduct my submission without retrainig, do the following steps:

1. Installation
2. Data Preparation
3. Training
4. Testing
5. Reference

Install Packages

• install pytorch from https://pytorch.org/get-started/locally/
• install dependencies

pip install -r requirements.txt

Data Preparation

Download the given dataset from Google Drive.

data /
  +- train_images
  +- test_images
  +- pascal_train.json
  +- test.json

Training

python train.py

※ get more info by python train.py --help

usage: train.py [-h] [-r WORKING_DIR] [-e EPOCH] [-b BATCH_SIZE]
                [-lr LEARNING_RATE]

optional arguments:
  -h, --help            show this help message and exit
  -r WORKING_DIR, --root WORKING_DIR
                        path to dataset
  -e EPOCH, --epochs EPOCH
                        num of epoch
  -b BATCH_SIZE, --batch BATCH_SIZE
                        set batch size
  -lr LEARNING_RATE, --learning_rate LEARNING_RATE
                        set learning rate

Testing

python test.py

Reference

• TorchVision Object Detection Finetuning Tutorial