pvrcnn_mot.yaml

# dataset info
dataset_path: "F:/data/tracking/training"
detections_path: "data/pvrcnn/training"
save_path: 'evaluation/results/sha_key/data'

# val [1,6,8,10,12,13,14,15,16,18,19]
# trainval [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20]
# test [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28]
tracking_seqs: [1,6,8,10,12,13,14,15,16,18,19]

# tracking type
tracking_type: "Car"

# KF parameters
state_func_covariance: 100
measure_func_covariance: 0.001
prediction_score_decay: 0.03
LiDAR_scanning_frequency: 10

# max prediction number of state function
max_prediction_num: 12
max_prediction_num_for_new_object: 2

# detection score threshold
input_score: 0
init_score: 2
update_score: 0
post_score: 0

# tracking latency (s)
# -1: global tracking
# 0.->500: online or near online tracking
latency: 0