search

Beniko95J / MLF-VO

Multi-Layer Fusion Visual Odometry
GNU General Public License v3.0
20 stars 3 forks source link

The result of evaluation,thank you ! #12

Open dixndnkx opened 1 month ago

dixndnkx commented 1 month ago

thank a lot for your work! I evaluate the model (RGB+Depth_multi_layer) using 3070,but the final result can t reach the result in paper.Here are my steps.Can you give me some suggestion? thanks!

python .\run_odometry.py -c .\configs\run_odometry.yaml run_odometry.yaml 72df2d39415684e4e7288731960f10f Then I use eval_odom.py to evaluate python odometry/eval_odom.py --result ./data/outputs/09 --align 7dof --seqs 09 --gt ./data/gt python odometry/eval_odom.py --result ./data/outputs/09 --align scale --seqs 09 --gt ./data/gt image Here is the result-7dof. sequence_09 Sequence: 09 Trans. err. (%): 4.627 Rot. err. (deg/100m): 0.365 ATE (m): 45.984 RPE (m): 0.074 RPE (deg): 0.415

The result-scale . sequence_09 Sequence: 09 Trans. err. (%): 5.269 Rot. err. (deg/100m): 0.365 ATE (m): 45.729 RPE (m): 0.081 RPE (deg): 0.415