Questions about the pretrained model and the trained model

[wennycooper]

Hi, thanks for the great paper.

I'm trying to reproduce the result in Table.1 in the paper. The 1st question is, is the model under the "pretrained_model" folder the best trained model? Or can you provide the trained best model?

I just run the command.sh with mode "test" as below:

python main.py \
    --mode test \
    --gpu 0 \
    --model pwclo_model \
    --data_root /mnt/SardinaDatasets/kitti_odom/dataset/sequences \
    --checkpoint_path ./pretrained_model/pretrained_model.ckpt \
    --log_dir test_log \
    --result_dir result \
    --train_list 0 1 2 3 4 5 6 \
    --val_list 7 8 9 10 \
    --test_list 0 1 2 3 4 5 6 7 8 9 10 \
    --num_H_input 64 \
    --num_W_input 1800 \
    --max_epoch 1000 \
    --learning_rate 0.001 \
    --batch_size 8 \
    > Efficient-LOnet_log.txt 2>&1 &

And in the output log file "Efficient-LOnet_log.txt", I did see some translation errors and rotation errors in some sequence. But why isn't all the sequence?

$ grep "Average" Efficient-LOnet_log.txt 
seq03 Average_t_error 11921.01 Average_r_error 71.75
seq04 Average_t_error 189.42 Average_r_error 81.33
seq09 Average_t_error 94657335718527820655687364593582080.00 Average_r_error 47.04

Any comment will be appreciated.