YOLOX implemented by pytorch lightning, a simpler expression of pytorch
python train.py -c configs/yolox_s.yaml
data_dir
├── annotations
│ ├── train.json
│ ├── val.json
├── _train
│ ├── <training images>
├── _val
│ ├── <validation images>
The parameters are important
class LitYOLOX(LightningModule):
def __init__(self, cfgs)
def on_train_start(self)
def training_step(self, batch, batch_idx)
def validation_step(self, batch, batch_idx)
def validation_epoch_end(self, val_step_outputs)
def configure_optimizers(self)
def on_train_end(self)