Deep learning-based Computer Vision Models for PyTorch

Boda (보다) means to see in Korean. This library was inspired by 🤗 Transformers.

Get started

git clone https://github.com/unerue/boda.git && cd boda
conda env create -f environment.yml
conda activate boda
python setup.py install

from boda.models import YolactConfig, YolactModel, YolactLoss

config = YolactConfig(num_classes=80)
model = YolactModel(config)
criterion = YolactLoss()

outputs = model(images)
losses = criterion(outputs, targets)
print(losses)

Comparison

Model	State	Training	Inference	Original	Ours
Mask R-CNN	😡	❌	❌
PolarMask	😡	❌	❌
YOLACT	😆	✔️	✔️
SOLOv1	🙂	❌	✔️
SOLOv2	🙂	❌	✔️
CenterMask	😡	❌	❌
YOLACT EDGE	😡	❌	❌

Misc

Model	State	Training	Inference	Original	Ours
SSD	🙂	❌	✔️
Faster R-CNN	🙂	❌	✔️
FCOS	🙂	❌	✔️
Keypoint R-CNN	🙂	❌	✔️
YOLOv4	😡	❌	❌

Pretrained Model Configurations

Model	Config name	Status	Original	Ours
SSD	ssd-base	🙂
	ssd-512	😡
Faster R-CNN	faster-rcnn-base	🙂
	faster-rcnn-r101	😡
Mask R-CNN	mask-rcnn-base	😡
	mask-rcnn-r50	😡
Keypoint R-CNN	keypoint-rcnn-base	🙂
	keypoint-rcnn-mobile	😡
FCOS	fcos-base	🙂
PolarMask	polarmask-base	😡
YOLOv4	yolov4-base	😡
YOLACT	yolact-base	😆
	yolact-r101	😡
	yolact-r101-300	😡
	yolact-r101-700	😡
SOLOv1	solov1-base	🙂
	solov1-r101	😡
SOLOv2	solov2-base	😡
CenterMask	centermask-base	😡
YOLACT EDGE	yolact-edge-base	😡