🎯_ Rotatated airplane Detector based on EfficientDet _

This rotation airplane detector is implemented and modified based on pytorch implementing EfficientDet(horizontal) and you can find original paper here.

🎯My Experiment

Prepare airplane dataset for Task1-Detection with oriented bounding boxesDOTA dataset

• evaluation metrics

coefficient	Input Size	mAP(iou=.5)
D2	768 x 768	0.574

• loss curve

Totall loss = regression loss + clasification loss

Outputs for rotated airplane detector

• Correct detection for all airplanes

  

• Detection of airplanes along with some error

  

🎯Installation

1. Install requirements:

conda create -n Rotateairplanedetection python=3.9  
conda activate Rotateairplanedetection  
conda install pytorch==1.7.0 torchvision==0.8.0 torchaudio==0.7.0 cudatoolkit=11.0 -c pytorch
pip install -r path to /requirements.txt  

Note: If you meet some troubles about installing environment, you can see the check.txt for more details.  

2. Install skew iou module:

cd polyiou
sudo apt-get install swig
swig -c++ -python polyiou.i
python setup.py build_ext --inplace

3. Install rotation nms module:

cd utils/nms
make

🎯Train

train procces is based on this great repo

A. Prepare dataset

you should put your dataset into datasets folder and the file tree is below.

# dataset structure should be like this
datasets/
    -your_project_name/
        -train_set_name/
            -*.jpg
        -val_set_name/
            -*.jpg
        -annotations
            -instances_{train_set_name}.json
            -instances_{val_set_name}.json

# for example, coco2017
datasets/
    -coco2017/
        -train2017/
            -000000000001.jpg
            -000000000002.jpg
            -000000000003.jpg
        -val2017/
            -000000000004.jpg
            -000000000005.jpg
            -000000000006.jpg
        -annotations
            -instances_train2017.json
            -instances_val2017.json

B. Manual set project's hyper parameters

you should create your own yml file under projects folder. My experiment yml file is rotation_vehicles.yml.

# create a yml file {your_project_name}.yml under 'projects' folder
# modify it following 'coco.yml'

# for example
project_name: coco
train_set: train2017
val_set: val2017
num_gpus: 4  # 0 means using cpu, 1-N means using gpus

# mean and std in RGB order, actually this part should remain unchanged as long as your dataset is similar to coco.
mean: [0.485, 0.456, 0.406]
std: [0.229, 0.224, 0.225]

# this is coco anchors, change it if necessary
anchors_scales: '[2 ** 0, 2 ** (1.0 / 3.0), 2 ** (2.0 / 3.0)]'
anchors_ratios: '[(1.0, 1.0), (1.4, 0.7), (0.7, 1.4)]'

# objects from all labels from your dataset with the order from your annotations.
# its index must match your dataset's category_id.
# category_id is one_indexed,
# for example, index of 'car' here is 2, while category_id of is 3
obj_list: ['person', 'bicycle', 'car', ...]

C.1 Train rotation detector on a custom dataset from scratch

# train rotation efficientdet-d0 on a custom dataset
# with batchsize 32 and learning rate 5e-3

python train.py -c 0 -p your_project_name --train_batch_size 32 --lr 5e-3

C.2 Train rotation detector with pretrained weight file which is trained on horizontal datasets.

you should download the pretrained horizontal detector weight file and put this pth file into weights folder.

# train rotation efficientdet-d0 on a custom dataset with pretrained weights which is trained on horizontal datasets.
# with batchsize 32 and learning rate 5e-3 for 10 epoches

python train.py -c 0 -p your_project_name --train_batch_size 32 --lr 5e-3 --num_epochs 10 \
 --load_weights /path/to/your/weights/efficientdet-d0.pth

# with a pretrained weight file, you can even freeze the backbone and train heads only
# to speed up training and help convergence.

python train.py -c 0 -p your_project_name --train_batch_size 32 --lr 5e-3 --num_epochs 10 \
 --load_weights /path/to/your/weights/efficientdet-d0.pth \
 --head_only True

D. Early stopping a training session.

# while training, press Ctrl+c, the program will catch KeyboardInterrupt
# and stop training, save current checkpoint.

E. Resume Training

# let says you started a training session like this.

python train.py -c 0 -p your_project_name --train_batch_size 32 --lr 5e-3 \
 --load_weights /path/to/your/weights/efficientdet-d0.pth \
 --head_only True

# then you stopped it with a Ctrl+c, it exited with a checkpoint

# now you want to resume training from the last checkpoint
# simply set load_weights to 'last'

python train.py -c 0 -p your_project_name --train_batch_size 16 --lr 1e-3 \
 --load_weights last \
 --head_only False

🎯Evaluation

The evaluation stage is implemented based on the following paper

file structure of evaluation stage

evaluation/
    -gt_labels/
        -*.txt
    -result_classname
        -Task1_{category_name}.txt
    -batch_inference.py
    -eval.py
    -imgnamefile.txt
    -prepare.py

💡References

The main framework of this project is based on the following very informative repository. Yet-Another-EfficientDet-Pytorch