like is

Method Seq. 09 Seq. 10
ORB-SLAM (full) 0.014 ± 0.008 0.012 ± 0.011
ORB-SLAM (short) 0.064 ± 0.141 0.064 ± 0.130
Mean Odom. 0.032 ± 0.026 0.028 ± 0.023

is mean Average sequence translational RMSE (%) ± Average sequence rotational error (deg/m) ???
thank u first

line 194, in computeOverallErr
ave_t_err = t_err / seq_len
ZeroDivisionError: division by zero

I debuged a littie, found the following seq_err is error, I used my own data, the poses_gt and poses_result could be read successfully.

compute sequence errors

        seq_err = self.calcSequenceErrors(poses_gt, poses_result)

Thank you for your outstanding work. I want to ask a question: kitti odometry's gt_ pose is relative to cam0 at t0, and my data is as follows: Given the position (x, y, z) of the camera on the vehicle in the world coordinate system and the conversion matrix from the camera to the world coordinate system at each time, how can I calculate the pose of the camera at each time relative to the camera at t0? Looking forward to your reply

Thanks for your sharing. The evaluation results are really comprehensive! Here, I have some question . I have read some papers which use kitti odometry. The evaluation result they give was in type of as follows:
• t: average translational RMSE drift (%) on a length of 100–800 m.
• r: average rotational RMSE drift (°/100 m) on a length of 100–800 m.

The results you give was:
Average sequence translation RMSE (%): 6.2131
Average sequence rotation error (deg/m): 0.0170
I think the translation RMSE are under the same evaluation, however, rotation error may not. I wonder whether it's right if I directly multiply 0.0170 by 100?

您好！
我运行了您的代码，但是对于代码怎么评估算法的这个原理看不懂，看了KITTI的数据集的论文也没有，请问您可以大致讲一下吗？十分感谢！！

大佬，我试了下readme里的方法，发现不行，能指教一下改什么吗，谢谢大佬

python evaluation.py --result_dir=./data/ --eva_seqs=09_pred,10_pred
Traceback (most recent call last):
File "evaluation.py", line 17, in
import tools.transformations as tr
ImportError: No module named tools.transformations

I generate a pose.txt from my SLAM algorithm.

while this tools throw an error, while I test on another evaluation tools it works fine.

could you kindly tell me the reason?

ZZZZ@DeskPC:~/Desktop/visual_odometry/KITTI_odometry_evaluation_tool$ python3 evaluation.py --result_dir=./data/ --eva_seqs=09_pred
Traceback (most recent call last):
  File "evaluation.py", line 663, in <module>
    pose_eval.eval(toCameraCoord=args.toCameraCoord)   # set the value according to the predicted results
  File "evaluation.py", line 592, in eval
    seq_err = self.calcSequenceErrors(poses_gt, poses_result)
  File "evaluation.py", line 165, in calcSequenceErrors
    pose_error = np.dot(np.linalg.inv(pose_delta_result), pose_delta_gt)
  File "/usr/local/lib/python3.5/dist-packages/numpy/linalg/linalg.py", line 551, in inv
    ainv = _umath_linalg.inv(a, signature=signature, extobj=extobj)
  File "/usr/local/lib/python3.5/dist-packages/numpy/linalg/linalg.py", line 97, in _raise_linalgerror_singular
    raise LinAlgError("Singular matrix")
numpy.linalg.LinAlgError: Singular matrix

Thank you very much for your work.
but when I use this tool for drawing the trajectory, the direction of (x,y,z) seems not correct.


How can I solve this problem??
Whether I use this in the wang way??

When the laser SLAM algorithm is run, the odometer track obtained is in the radar coordinate system, and the ground_truth is under the camera coordinate system. During the trajectory evaluation, we can use the params --toCameraCoord True to transform the trajectory to the camera coordinate system.
But the external parameter matrix in the program is just to adjust the coordinate system
R_C2L = np.array([[0, 0, 1, 0],
[-1, 0, 0, 0],
[0, -1, 0, 0],
[0, 0, 0, 1]])
But KITTI's calibration file gives the transformation relationship from radar to camera coordinate system, Tr in calib.txt
Tr: 4.276802385584e-04 -9.999672484946e-01 -8.084491683471e-03 -1.198459927713e-02 -7.210626507497e-03 8.081198471645e-03 -9.999413164504e-01 -5.403984729748e-02 9.999738645903e-01 4.859485810390e-04 -7.206933692422e-03 -2.921968648686e-01

My question is, when I use this real external parameter, I get a larger error result, but using this default parameter, I can get a smaller error result. How can I evaluate the laser slam algorithm correctly?
The first evaluation uses the external parameters of the calibration file, and the second evaluation uses the default parameters in the program

I'm trying to evaluate KITTI odometry
first I convert pose to 6DoF using this code

def rotationMatrixToEulerAngles(self, R):
        assert (self.isRotationMatrix(R))
        sy = math.sqrt(R[0, 0] * R[0, 0] + R[1, 0] * R[1, 0])
        singular = sy < 1e-6

        if not singular:
            x = math.atan2(R[2, 1], R[2, 2])
            y = math.atan2(-R[2, 0], sy)
            z = math.atan2(R[1, 0], R[0, 0])
        else:
            x = math.atan2(-R[1, 2], R[1, 1])
            y = math.atan2(-R[2, 0], sy)
            z = 0
        return np.array([x, y, z], dtype=np.float32)

def matrix_rt(self, p):
        return np.vstack([np.reshape(p.astype(np.float32), (3, 4)), [[0., 0., 0., 1.]]])

pose1 = self.matrix_rt(self.poses[index][i])
pose2 = self.matrix_rt(self.poses[index][i + 1])
pose2wrt1 = np.dot(np.linalg.inv(pose1), pose2)
R = pose2wrt1[0:3, 0:3]
t = pose2wrt1[0:3, 3]
angles = self.rotationMatrixToEulerAngles(R)
odometries.append(np.concatenate((t, angles)))

also, model output is in the same format (6DoF)

the question is how to evaluate 6DoF results