2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2018
DOI: 10.1109/iros.2018.8593379
|View full text |Cite
|
Sign up to set email alerts
|

Unsupervised Trajectory Segmentation and Promoting of Multi-Modal Surgical Demonstrations

Abstract: To improve the efficiency of surgical trajectory segmentation for robot learning in robot-assisted minimally invasive surgery, this paper presents a fast unsupervised method using video and kinematic data, followed by a promoting procedure to address the over-segmentation issue. Unsupervised deep learning network, stacking convolutional auto-encoder, is employed to extract more discriminative features from videos in an effective way. To further improve the accuracy of segmentation, on one hand, wavelet transfo… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1
1

Citation Types

0
6
0

Year Published

2021
2021
2024
2024

Publication Types

Select...
4
2
1

Relationship

0
7

Authors

Journals

citations
Cited by 7 publications
(6 citation statements)
references
References 14 publications
0
6
0
Order By: Relevance
“…Luongo et al [37] achieve 71% mAP using only kinematic data as input to an RNN on the JIGSAWS dataset. Other works that include images as input data report results of 70.6% mAP [26] and 79.1% mAP [27]. Thus, we consider that we achieved a good result, especially taking into account that recognition is performed on tasks that do not follow a specific order in any of the attempts.…”
Section: Discussionmentioning
confidence: 80%
See 1 more Smart Citation
“…Luongo et al [37] achieve 71% mAP using only kinematic data as input to an RNN on the JIGSAWS dataset. Other works that include images as input data report results of 70.6% mAP [26] and 79.1% mAP [27]. Thus, we consider that we achieved a good result, especially taking into account that recognition is performed on tasks that do not follow a specific order in any of the attempts.…”
Section: Discussionmentioning
confidence: 80%
“…It involves breaking trajectories into sub-trajectories, facilitating learning from demonstrations, skill assessment, phase recognition, among other applications. Authors leverage kinematics information provided by surgical robots, which when combined with video data, yields improved accuracy results [27] [28].…”
Section: Introductionmentioning
confidence: 99%
“…This work is based on a structure of dense connection, in which the first half of the network, the dense block, is an encoder that performs feature extraction, the transition layer performs the trajectory segmentation, and the up-sampling layer is used for image reconstruction. A similar approach is presented in [77], but using a compact stacking convolutional autoencoder model and wavelet transform based filtering. Marban et al [79] propose a method to estimate the position and velocity of the instruments in 3D from monocular videos using a regression model based on CNN+LSTM.…”
Section: ) Trajectory Segmentationmentioning
confidence: 99%
“…It involves breaking trajectories into sub-trajectories, facilitating learning from demonstrations, skill assessment, phase recognition, among other applications. Authors take advantage of kinematics information provided by surgical robots, which combined with video data, allows more accurate results [26,27].…”
Section: Introductionmentioning
confidence: 99%