Enhancing Geometric Factors in Model Learning and Inference for Object Detection and Instance Segmentation

Zheng, Zhaohui; Wang, Ping; Ren, Dongwei; Liu, Wei; Ye, Rongguang; Hu, Qinghua; Zuo, Wangmeng

doi:10.48550/arxiv.2005.03572

Cited by 19 publications

(19 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The algorithm adds an adaptive anchor boxes calculation at the input end, making it easy to find the best anchor boxes to improve the model's ability to locate cells. 4) CIoU loss [16]. In the labeled samples of the bone marrow cell detection task, the rectangular boxes of the labeled cells are often unable to frame the cells due to manual labeling accurately.…”

Section: A Effective Core Modulementioning

confidence: 99%

Bone Marrow Cell Recognition: Training Deep Object Detection with A New Loss Function

Huang¹,

Cheng²,

Fan³

et al. 2021

Preprint

View full text Add to dashboard Cite

For a long time, bone marrow cell morphology examination has been an essential tool for diagnosing blood diseases. However, it is still mainly dependent on the subjective diagnosis of experienced doctors, and there is no objective quantitative standard. Therefore, it is crucial to study a robust bone marrow cell detection algorithm for a quantitative automatic analysis system. Currently, due to the dense distribution of cells in the bone marrow smear and the diverse cell classes, the detection of bone marrow cells is difficult. The existing bone marrow cell detection algorithms are still insufficient for the automatic analysis system of bone marrow smears. This paper proposes a bone marrow cell detection algorithm based on the YOLOv5 network, trained by minimizing a novel loss function. The classification method of bone marrow cell detection tasks is the basis of the proposed novel loss function. Since bone marrow cells are classified according to series and stages, part of the classes in adjacent stages are similar. The proposed novel loss function considers the similarity between bone marrow cell classes, increases the penalty for prediction errors between dissimilar classes, and reduces the penalty for prediction errors between similar classes. The results show that the proposed loss function effectively improves the algorithm's performance, and the proposed bone marrow cell detection algorithm has achieved better performance than other cell detection algorithms.

show abstract

Section: A Effective Core Modulementioning

confidence: 99%

Bone Marrow Cell Recognition: Training Deep Object Detection with A New Loss Function

Huang¹,

Cheng²,

Fan³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…In order to solve the issue of IoU when considering it as a loss function, several alternative formulations were suggested in the literature, e.g. (Rezatofighi et al 2019) proposed the Generalized IoU (GIoU) loss, (Zheng et al 2020a) proposed the Distance IoU (DIoU) loss, while only recently (Zheng et al 2020b) proposed the Complete IoU (CIoU) loss, which has shown promising results and faster convergence than GIoU and DIoU. It is defined as:…”

Section: Complete Intersection Over Union (Ciou)mentioning

confidence: 99%

“…Regarding the bounding boxes coordinates refinement, differently from previous works that use the Smooth L1 loss, our model adopts the Complete IoU loss (Zheng et al 2020b). To the best of our knowledge, this is the first work adopting the Complete IOU loss in order to refine the final bounding boxes coordinates.…”

Section: Trainingmentioning

confidence: 99%

“…Regarding the parameter alpha in Eq. 5, we just used the value specified in (Zheng et al 2020b) which is identified by a specific formula.…”

Section: Implementation Detailsmentioning

confidence: 99%

“…• we present a new grounding loss for visual proposals which considers also the object proposals semantic information, differently from the works in the literature which just consider their shapes and spatial positions in the image; • we are the first to adopt the Complete Intersection over Union (Zheng et al 2020b) loss for the visual grounding task; • we introduce a new regression loss on the proposal bounding boxes coordinates which is applied to a subset of all the proposals, selected by considering the object proposals semantic information. This loss differs from the one used by the approaches in the literature, which only considers the proposal with the largest overlap with the ground truth.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A Better Loss for Visual-Textual Grounding

Rigoni,

Serafini,

Sperduti

2021

Preprint

View full text Add to dashboard Cite

Given a textual phrase and an image, the visual grounding problem is defined as the task of locating the content of the image referenced by the sentence. It is a challenging task that has several real-world applications in human-computer interaction, image-text reference resolution, and video-text reference resolution. In the last years, several works have addressed this problem with heavy and complex models that try to capture visual-textual dependencies better than before. These models are typically constituted by two main components that focus on how to learn useful multi-modal features for grounding and how to improve the predicted bounding box of the visual mention, respectively. Finding the right learning balance between these two sub-tasks is not easy, and the current models are not necessarily optimal with respect to this issue. In this work, we propose a model that, although using a simple multi-modal feature fusion component, is able to achieve a higher accuracy than state-of-the-art models thanks to the adoption of a more effective loss function, based on the classes probabilities, that reaches, in the considered datasets, a better learning balance between the two sub-tasks mentioned above.

show abstract

Automated detection of GFAP-labeled astrocytes in micrographs using YOLOv5

Huang

Kruyer

Syed

et al. 2022

Sci Rep

View full text Add to dashboard Cite

Astrocytes, a subtype of glial cells with a complex morphological structure, are active players in many aspects of the physiology of the central nervous system (CNS). However, due to their highly involved interaction with other cells in the CNS, made possible by their morphological complexity, the precise mechanisms regulating astrocyte function within the CNS are still poorly understood. This knowledge gap is also due to the current limitations of existing quantitative image analysis tools that are unable to detect and analyze images of astrocyte with sufficient accuracy and efficiency. To address this need, we introduce a new deep learning framework for the automated detection of GFAP-immunolabeled astrocytes in brightfield or fluorescent micrographs. A major novelty of our approach is the applications of YOLOv5, a sophisticated deep learning platform designed for object detection, that we customized to derive optimized classification models for the task of astrocyte detection. Extensive numerical experiments using multiple image datasets show that our method performs very competitively against both conventional and state-of-the-art methods, including the case of images where astrocytes are very dense. In the spirit of reproducible research, our numerical code and annotated data are released open source and freely available to the scientific community.

show abstract

Enhancing Geometric Factors in Model Learning and Inference for Object Detection and Instance Segmentation

Cited by 19 publications

References 51 publications

Bone Marrow Cell Recognition: Training Deep Object Detection with A New Loss Function

Bone Marrow Cell Recognition: Training Deep Object Detection with A New Loss Function

A Better Loss for Visual-Textual Grounding

Automated detection of GFAP-labeled astrocytes in micrographs using YOLOv5

Contact Info

Product

Resources

About