A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
Filters
Multi-Prompt Fine-Tuning of Foundation Models for Enhanced Medical Image Segmentation
[article]
2023
arXiv
pre-print
Preserved Fulltext
The Segment Anything Model (SAM) is a powerful foundation model that introduced revolutionary advancements in natural image segmentation. ...
wide range of segmentation tasks. ...
The Segment Anything Model (SAM), introduced by the SA project, enjoys robust zero-shot performance comparable to many supervised methods [9] . ...
arXiv:2310.02381v1
fatcat:h2ao64opqnbipoxudwmay6vkr4
A Survey on Segment Anything Model (SAM): Vision Foundation Model Meets Prompt Engineering
[article]
2023
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
Other Versions
Trained on a large segmentation dataset of over 1 billion masks, SAM is capable of segmenting any object on a certain image. ...
Segment anything model (SAM) developed by Meta AI Research has recently attracted significant attention. ...
Another recent work performs comprehensive evaluations on the robustness of SAM on corruption and beyond. ...
arXiv:2306.06211v3
fatcat:wtfxuprg25akdobdtx7u54huwu
Polyp-SAM++: Can A Text Guided SAM Perform Better for Polyp Segmentation?
[article]
2023
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
We will evaluate the performance of a text-guided SAM on the polyp segmentation task on benchmark datasets. We will also compare the results of text-guided SAM vs unprompted SAM. ...
In the field of medical image segmentation, polyp segmentation holds a position of high importance, thus creating a model which is robust and precise is quite challenging. ...
lang-segment-anything. ...
arXiv:2308.06623v1
fatcat:zmrpxyhg2feqxix5rc6vlvee5y
Robustness of SAM: Segment Anything Under Corruptions and Beyond
[article]
2023
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
Note that this fulltext copy is not of the "primary" version of this work. The version it corresponds to is:
2023 pre-print arXiv:2306.07713v1 fatcat:7bd3mqxkdvcabiino3bfvjzvuuOther Versions
Segment anything model (SAM), as the name suggests, is claimed to be capable of cutting out any object and demonstrates impressive zero-shot transfer performance with the guidance of prompts. ...
To the best of our knowledge, our work is the first of its kind to evaluate the robustness of SAM under style change, local occlusion, and local adversarial patch attacks. ...
Conclusions In this work, we are among the early pioneers to evaluate the robustness of the segment anything model (SAM), for which we provide a comprehensive evaluation. ...
arXiv:2306.07713v3
fatcat:vw3fsccx4reb7ihli6vigltfgi
AquaSAM: Underwater Image Foreground Segmentation
[article]
2023
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
The Segment Anything Model (SAM) has revolutionized natural image segmentation, nevertheless, its performance on underwater images is still restricted. ...
This work presents AquaSAM, the first attempt to extend the success of SAM on underwater images with the purpose of creating a versatile method for the segmentation of various underwater targets. ...
of the Dice loss and cross-entropy loss, which has demonstrated robustness in various segmentation tasks [30] [31] . ...
arXiv:2308.04218v1
fatcat:lh7w2vzu3ratplxdd3vkwhi6fy
CEmb-SAM: Segment Anything Model with Condition Embedding for Joint Learning from Heterogeneous Datasets
[article]
2023
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2024; you can also visit the original URL.
The file type is application/pdf.
For robust segmentation, we leverage recently proposed Segment Anything model (SAM) in order to incorporate sub-group information into the model. ...
Although using the common modality of ultrasound, one typically needs separate datasets in order to segment, for example, different anatomical structures or lesions with different levels of malignancy. ...
Our main contributions are as follows: -We propose CEmb-SAM, which jointly trains a model over heterogeneous datasets leveraging Segment Anything model for robust segmentation performances. ...
arXiv:2308.06957v1
fatcat:l26p3b2wy5acdmtbp7we44g6tu
SSR: SAM is a Strong Regularizer for domain adaptive semantic segmentation
[article]
2024
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2024; you can also visit the original URL.
The file type is application/pdf.
We introduced SSR, which utilizes SAM (segment-anything) as a strong regularizer during training, to greatly enhance the robustness of the image encoder for handling various domains. ...
Meanwhile, the ImageNet pre-trained image encoder is still a mature choice of backbone for the semantic segmentation task, especially when the SAM is category-irrelevant. ...
SAM (Segment-anything [33] ) is a vision foundation model trained on a massive-scale dataset (SA-1B). ...
arXiv:2401.14686v1
fatcat:hpm532rtvrc4bkdtzt3flnsega
SAM Meets Robotic Surgery: An Empirical Study on Generalization, Robustness and Adaptation
[article]
2023
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
The Segment Anything Model (SAM) serves as a fundamental model for semantic segmentation and demonstrates remarkable generalization capabilities across a wide range of downstream scenarios. ...
In this empirical study, we examine SAM's robustness and zero-shot generalizability in the field of robotic surgery. ...
https://github.com/hendrycks/robustness
https://github.com/facebookresearch/segment-anything ...
arXiv:2308.07156v1
fatcat:hrlkpkiyvbds7n5iir53tyyuqa
From Generalization to Precision: Exploring SAM for Tool Segmentation in Surgical Environments
[article]
2024
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2024; you can also visit the original URL.
The file type is application/pdf.
Initial exploratory works with the Segment Anything Model (SAM) show that bounding-box-based prompting presents notable zero-short generalization. ...
Method: We use SAM to generate the over-segmented prediction of endoscopic frames. ...
Note that to test the zero-shot generalizability of the network, the segment anything model was used out-of-the-box with its predefined weights, with no additional fine-tuning to the datasets. ...
arXiv:2402.17972v1
fatcat:ztrdlbfrr5cubhfawbkezi3tme
Personalize Segment Anything Model with One Shot
[article]
2023
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
Note that this fulltext copy is not of the "primary" version of this work. The version it corresponds to is:
2023 pre-print arXiv:2305.03048v1 fatcat:bnfrt52u2rbnvcenitukywdjdm
Driven by large-data pre-training, Segment Anything Model (SAM) has been demonstrated as a powerful and promptable framework, revolutionizing the segmentation models. ...
To further alleviate the mask ambiguity, we present an efficient one-shot fine-tuning variant, PerSAM-F. ...
Overall, our PerSAM-F indicates better robustness to quality of the given mask than SegGPT. 4 . Robustness to Quality of the One-shot Mask. ...
arXiv:2305.03048v2
fatcat:uv5pdhefyrdgdcx4ekl6dseanu
Grounded SAM: Assembling Open-World Models for Diverse Visual Tasks
[article]
2024
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2024; you can also visit the original URL.
The file type is application/pdf.
Grounded SAM also shows superior performance on open-vocabulary benchmarks, achieving 48.7 mean AP on SegInW (Segmentation in the wild) zero-shot benchmark with the combination of Grounding DINO-Base and ...
This integration enables the detection and segmentation of any regions based on arbitrary text inputs and opens a door to connecting various vision models. ...
By leveraging the capabilities of these two robust expert models, the open-set detection and segmentation tasks can be more effortlessly accomplished. ...
arXiv:2401.14159v1
fatcat:hh2oaburnffjnemzpld564qynu
Matte Anything: Interactive Natural Image Matting with Segment Anything Models
[article]
2024
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2024; you can also visit the original URL.
The file type is application/pdf.
Note that this fulltext copy is not of the "primary" version of this work. The version it corresponds to is:
2023 pre-print arXiv:2306.04121v1 fatcat:2gbaziq42rgajdmxjvwjziyg2m
Specifically, we use the segment anything model to predict high-quality contour with user interaction and an open-vocabulary detector to predict the transparency of any object. ...
However, the production of trimap often requires significant labor, which limits the widespread application of matting algorithms on a large scale. ...
Recently, Kirillov et al. have introduced the Segment Anything Model (SAM) [16] as a segmentation foundation model in computer vision, capable of segmenting any object based on user prompts. ...
arXiv:2306.04121v2
fatcat:kac5v5mg55e67f7jullcfbmulm
Black-box Targeted Adversarial Attack on Segment Anything (SAM)
[article]
2024
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2024; you can also visit the original URL.
The file type is application/pdf.
Other Versions
Realizing flexible attacks on SAM is beneficial for understanding the robustness of SAM in the adversarial context. To this end, this work aims to achieve a targeted adversarial attack (TAA) on SAM. ...
Recently, Segment Anything Model (SAM) has emerged to become a popular foundation model in computer vision due to its impressive generalization to unseen data and tasks. ...
Beyond the segmentation object, SAM is also used for image editing, such as Magic Copy Kevmo ( 2023 ), which focuses on extracting the foreground using the capability of segmenting anything of SAM. ...
arXiv:2310.10010v2
fatcat:zdumbff2ejdzjhnbpwcpbd4qgm
Segment Anything Meets Point Tracking
[article]
2023
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
Note that this fulltext copy is not of the "primary" version of this work. The version it corresponds to is:
2023 pre-print arXiv:2307.01197v1 fatcat:fvyzokmwsncxbjnmfdcbb2vw6q
We highlight the merits of point-based tracking through direct evaluation on the zero-shot open-world Unidentified Video Objects (UVO) benchmark. ...
The Segment Anything Model (SAM) has established itself as a powerful zero-shot image segmentation model, enabled by efficient point-centric annotation and prompt-based models. ...
One such promising model is the Segment Anything Model (SAM) [16] . ...
arXiv:2307.01197v2
fatcat:dftcss2whzdela7v7m4mqhnigi
Part-aware Personalized Segment Anything Model for Patient-Specific Segmentation
[article]
2024
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2024; you can also visit the original URL.
The file type is application/pdf.
To further promote the robustness of the selected prompt, we propose a retrieval approach to handle outlier prompts. ...
We introduce a novel part-aware prompt mechanism to select multiple-point prompts based on part-level features of the one-shot data. ...
Adapt SAM to Medical Image Domain with Fine-tuning Segment Anything Model (SAM) [24] is initially pre-trained on the SA-1B [24] dataset in the natural image domain. ...
arXiv:2403.05433v1
fatcat:h3abokhykvdppp2prfcxoqylne
« Previous
Showing results 1 — 15 out of 66,142 results