config.yaml

BACKBONE_NAME: efficientnetb2
BATCH_SIZE: 2
CENTERNET_PRETRAIN_PATH: null
CONFIDENCE_THRESHOLD: 0.1
CP_SAVING_PATH: ./Dataset/checkpoints/efficientnetb2_CenterNet.hdf5
DEPTH_LOSS_WEIGHT: 1.5
DUPLICATE_CAR_DISTANCE_THRESHOLD: 15
EPOCH: 16
HEATMAP_LOSS_WEIGHT: 0.1
HFLIP_FOR_TRAIN: true
HFLIP_PROB: 0.5
INPUT_IMAGE_SIZE: !!python/tuple
- 512
- 1536
LEARNING_RATE: 0.0001
MAIN_DIR: ./Dataset/pku-autonomous-driving/
MAX_OBJECTS: 50
ORIGINAL_IMAGE_SIZE: !!python/tuple
- 2710
- 3384
ORIGINAL_Y_CROPPED: 1526
OUTPUT_FILE_PATH: ./OutputFile/
OUTPUT_SIZE: !!python/tuple
- 128
- 384
ROTATE_FOR_TRAIN: true
ROTATE_PROB: 0.5
ROTATION_LOSS_WEIGHT: 1.25
SAMPLE_SUBMISSION_CSV_DIR: ./Dataset/pku-autonomous-driving/sample_submission.csv
SIGMA_BASE: 1
TEST_IMAGE_DIR: ./Dataset/pku-autonomous-driving/test_images
TEST_MASK_DIR: ./Dataset/pku-autonomous-driving/test_masks
TRAIN_CSV_DIR: ./Dataset/pku-autonomous-driving/train.csv
TRAIN_IMAGE_DIR: ./Dataset/pku-autonomous-driving/train_images
TRAIN_MASK_DIR: ./Dataset/pku-autonomous-driving/train_masks
X_ROT_ANGLE: 4
Y_ROT_ANGLE: 4
Z_ROT_ANGLE: 4