List of Papers By topics Author List
| Paper Info | Reviews | Meta-review | Author Feedback | Post-Rebuttal Meta-reviews |
Authors
Qilong Zhangli, Jingru Yi, Di Liu, Xiaoxiao He, Zhaoyang Xia, Qi Chang, Ligong Han, Yunhe Gao, Song Wen, Haiming Tang, He Wang, Mu Zhou, Dimitris Metaxas
Abstract
Top-down instance segmentation framework has shown its superiority in object detection compared to the bottom-up framework. While it is efficient in addressing over-segmentation, top-down instance segmentation suffers from over-crop problem. However, a complete segmentation mask is crucial for biological image analysis as it delivers important morphological properties such as shapes and volumes. In this paper, we propose a region proposal rectification (RPR) module to address this challenging incomplete segmentation problem. In particular, we offer a progressive ROIAlign module to introduce neighbor information into a series of ROIs gradually. The ROI features are fed into an attentive feed-forward network (FFN) for proposal box regression. With additional neighbor information, the proposed RPR module shows significant improvement in correction of region proposal locations and thereby exhibits favorable instance segmentation performances on three biological image datasets compared to state-of-the-art baseline methods. Experimental results demonstrate that the proposed RPR module is effective in both anchor-based and anchor-free top-down instance segmentation approaches, suggesting the proposed method can be applied to general top-down instance segmentation of biological images.
Link to paper
DOI: https://link.springer.com/chapter/10.1007/978-3-031-16440-8_13
SharedIt: https://rdcu.be/cVRvx
Link to the code repository
https://github.com/qzhangli/RPR
Link to the dataset(s)
https://www.kaggle.com/c/data-science-bowl-2018
https://www.plant-phenotyping.org/datasets-home
Reviews
Review #2
- Please describe the contribution of the paper
- This paper proposes a region proposal rectification (RPR) module which involves two components: a progressive ROIAlign and an attentive feed-forward network (FFN).
- RPR shows improvement in region proposal location rectification and achieves favorable performances in instance segmentation for both anchor-based and anchor-free approaches (e.g., Mask R-CNN and CenterMask) in three different biological image datasets.
- Please list the main strengths of the paper; you should write about a novel formulation, an original way to use data, demonstration of clinical feasibility, a novel application, a particularly strong evaluation, or anything else that is a strong aspect of this work. Please provide details, for instance, if a method is novel, explain what aspect is novel and why this is interesting.
- The paper is well organized and easy to follow.
- The proposed region proposal rectification (RPR) is well motivated and improves performances for both anchor-based and anchor-free approaches.
- Please list the main weaknesses of the paper. Please provide details, for instance, if you think a method is not novel, explain why and provide a reference to prior work.
- I am both fine with the methodology part and the empirical results part. This paper has some novelty and provides new perspectives to redefine the procedure of biological instance mask generation. It’s relative good and solid application paper, and I don’t see much weakness.
- Please rate the clarity and organization of this paper
Good
- Please comment on the reproducibility of the paper. Note, that authors have filled out a reproducibility checklist upon submission. Please be aware that authors are not required to meet all criteria on the checklist - for instance, providing code and data is a plus, but not a requirement for acceptance
This paper is reproducible based on the detailed descriptions of the proposed method.
- Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review: https://conferences.miccai.org/2022/en/REVIEWER-GUIDELINES.html
See the details above.
- Rate the paper on a scale of 1-8, 8 being the strongest (8-5: accept; 4-1: reject). Spreading the score helps create a distribution for decision-making
6
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
This paper looks interesting and achieves good results.
- Number of papers in your stack
5
- What is the ranking of this paper in your review stack?
1
- Reviewer confidence
Confident but not absolutely certain
- [Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed
N/A
- [Post rebuttal] Please justify your decision
N/A
Review #3
- Please describe the contribution of the paper
This work tackles the incomplete instance segmentation using existing methods such as Mask RCNN by proposing a novel region proposal rectification (RPR) module that rectifies the region proposal locations with an expanded view.
- Please list the main strengths of the paper; you should write about a novel formulation, an original way to use data, demonstration of clinical feasibility, a novel application, a particularly strong evaluation, or anything else that is a strong aspect of this work. Please provide details, for instance, if a method is novel, explain what aspect is novel and why this is interesting.
This work is well motivated with a clear problem setting and analysis of existing works. The method development is lucid and easy to follow with a nice method diagram. The improvement of the proposed method in the experiments is impressive.
- Please list the main weaknesses of the paper. Please provide details, for instance, if you think a method is not novel, explain why and provide a reference to prior work.
- No evaluation experiment on the hyperparameter K. I guess for different datasets, the best K would be different.
- How many parameters and memory consumption are introduced for the attentive FFN? Since K is relatively small, I guess the memory consumption would be ignorable. However, these details are still expected to be included.
- If possible, add a section about related works.
- Please rate the clarity and organization of this paper
Very Good
- Please comment on the reproducibility of the paper. Note, that authors have filled out a reproducibility checklist upon submission. Please be aware that authors are not required to meet all criteria on the checklist - for instance, providing code and data is a plus, but not a requirement for acceptance
The key idea of this work is simple. Though no code is released, it wouldn’t be difficult to reproduce the results on the public datasets.
- Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review: https://conferences.miccai.org/2022/en/REVIEWER-GUIDELINES.html
- The weakness I mentioned above might be helpful for the authors to improve their work.
- I would suggest the authors test on more challenging datasets and if possible give some failure cases of this method. As shown in Table 1, the proposed method does not beat the baseline on every metric. It’s interesting to see and analyze some failure cases.
- Rate the paper on a scale of 1-8, 8 being the strongest (8-5: accept; 4-1: reject). Spreading the score helps create a distribution for decision-making
6
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
This work is well-written with clear motivation, novel method, and impressive results. The limitations of this work do not hurt the overall quality of this work.
- Number of papers in your stack
5
- What is the ranking of this paper in your review stack?
1
- Reviewer confidence
Confident but not absolutely certain
- [Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed
N/A
- [Post rebuttal] Please justify your decision
N/A
Review #4
- Please describe the contribution of the paper
The authors propose a region proposal rectification module, which includes a progressive ROIAlign module and a self-similarity attention based feed forward network module, to address the issue of bounding box not enclosing the entire object in object detection.
- Please list the main strengths of the paper; you should write about a novel formulation, an original way to use data, demonstration of clinical feasibility, a novel application, a particularly strong evaluation, or anything else that is a strong aspect of this work. Please provide details, for instance, if a method is novel, explain what aspect is novel and why this is interesting.
The proposed module RPR module enriches the features for bounding box regression by progressively looking at the neighboring areas of the initial proposal and a self-similarity based attention module.The module was evaluated on three datasets using objection detection networks with anchor-based and anchor-free region proposal networks.
- Please list the main weaknesses of the paper. Please provide details, for instance, if you think a method is not novel, explain why and provide a reference to prior work.
The rationale for RPR module was to improve volume and shape quantification. However from the results in Table 1, the ASD improvement for segmentation is lower than that of bbox regression. Only average precision is used as the evaluation metric. Dice coefficient and average surface distance for segmentation task might provide additional information on the value of RPR module.
- Please rate the clarity and organization of this paper
Very Good
- Please comment on the reproducibility of the paper. Note, that authors have filled out a reproducibility checklist upon submission. Please be aware that authors are not required to meet all criteria on the checklist - for instance, providing code and data is a plus, but not a requirement for acceptance
Sufficient detail provided on the module and the experiments to evaluate the approach.
- Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review: https://conferences.miccai.org/2022/en/REVIEWER-GUIDELINES.html
The authors propose an RPR module to improve object detection and thereby improve segmentation quality for better volume and shape estimation. The paper is well written.
- Given that the improvements expected are small, mainly along excluded areas near the boundary of the object, average precision as the only evaluation might be insufficient. Consider including more metrics like Dice coefficient and Average surface distance / Hausdorff distance.
- The improvement in AP for segmentation task when using mask RCNN with RPR was <2% in Table 1. However the improvement in IoU in Fig 4a is >20%. Please provide the IoU plot after non max suppression.
- From Fig.5,the proposed module improves the regression of bounding boxes but the segmentation AP improves only marginally. Suggest looking into other segmentation metrics to understand if the proposed module improves sensitivity. Also, the edges could be weak, making it hard to segment; i.e., good bbox regression might not always improve the segmentation quality.
- Rate the paper on a scale of 1-8, 8 being the strongest (8-5: accept; 4-1: reject). Spreading the score helps create a distribution for decision-making
5
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
The proposed module is novel and improves bbox regression. However, the segmentation quality improves marginally.
- Number of papers in your stack
7
- What is the ranking of this paper in your review stack?
4
- Reviewer confidence
Confident but not absolutely certain
- [Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed
N/A
- [Post rebuttal] Please justify your decision
N/A
Primary Meta-Review
- Please provide your assessment of this work, taking into account all reviews. Summarize the key strengths and weaknesses of the paper and justify your recommendation. In case you deviate from the reviewers’ recommendations, explain in detail the reasons why. In case of an invitation for rebuttal, clarify which points are important to address in the rebuttal.
Reviewers concur on the adequacy of the technical novelty and the quality of results, recommending acceptance unanimously. The final version should include all reviewers’ comments and suggestions. In particular: (1) Additional experiments (R2), (2) complexity analysis in the results (R2), and (3) Additional performance metrics and discussions (R3).
- What is the ranking of this paper in your stack? Use a number between 1 (best paper in your stack) and n (worst paper in your stack of n papers). If this paper is among the bottom 30% of your stack, feel free to use NR (not ranked).
2
Author Feedback
Thank you for all the valuable comments from reviewers and meta-reviewer. We have carefully revised the paper and made updates to the questions accordingly. We will have those updates in the final version.