Wasserstein Loss based Deep Object Detection

Han, Yuzhuo; Liu, Xiaofeng; Sheng, Zhenfei; Ren, Yutao; Han, Xu; You, Jane; Liu, Risheng; Luo, Zhongxuan

doi:10.1109/cvprw50498.2020.00507

Cited by 29 publications

(16 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Focal loss [22] did not work in their case. Recently, a Wasserstein loss was proposed for weighting detector errors by severity [14], similarly to the Earth Mover's Distance used by [15] and only applied to a softmax-based Faster-RCNN [27]. We show that hierarchical training can indeed be done using focal loss.…”

Section: Benefit Deep Featuresmentioning

confidence: 94%

United We Learn Better: Harvesting Learning Improvements From Class Hierarchies Across Tasks

Shkodrani¹,

Wang²,

Manfredi³

et al. 2021

Preprint

View full text Add to dashboard Cite

Attempts of learning from hierarchical taxonomies in computer vision have been mostly focusing on image classification. Though ways of best harvesting learning improvements from hierarchies in classification are far from being solved, there is a need to target these problems in other vision tasks such as object detection. As progress on the classification side is often dependent on hierarchical crossentropy losses, novel detection architectures using sigmoid as an output function instead of softmax cannot easily apply these advances, requiring novel methods in detection. In this work we establish a theoretical framework based on probability and set theory for extracting parent predictions and a hierarchical loss that can be used across tasks, showing results across classification and detection benchmarks and opening up the possibility of hierarchical learning for sigmoid-based detection architectures. 1

show abstract

Section: Benefit Deep Featuresmentioning

confidence: 94%

United We Learn Better: Harvesting Learning Improvements From Class Hierarchies Across Tasks

Shkodrani¹,

Wang²,

Manfredi³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…With the low-cost modification of the loss function perspective, our solution can be added on any up-to-date general deep networks in a plug-and-play fashion. The distance is defined as the cost of optimal transport for moving the mass in one distribution to match the target distribution [13,14]. Specifically, we measure the discrete optimal transport distance between a softmax prediction and its target label, both of which are normalized as histograms.…”

Section: … …mentioning

confidence: 99%

Embedding Semantic Hierarchy in Discrete Optimal Transport for Risk Minimization

et al. 2021

ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

The widely-used cross-entropy (CE) loss-based deep networks achieved significant progress w.r.t. the classification accuracy. However, the CE loss can essentially ignore the risk of misclassification which is usually measured by the distance between the prediction and label in a semantic hierarchical tree. In this paper, we propose to incorporate the risk-aware inter-class correlation in a discrete optimal transport (DOT) training framework by configuring its ground distance matrix. The ground distance matrix can be pre-defined following a priori of hierarchical semantic risk. Specifically, we define the tree induced error (TIE) on a hierarchical semantic tree and extend it to its increasing function from the optimization perspective. The semantic similarity in each level of a tree is integrated with the information gain. We achieve promising results on several large scale image classification tasks with a semantic tree structure in a plug and play manner.

show abstract

“…In recent years, big data drives the fast development of deep learning, which has transformed many fields, such as computer vision and medical image analysis (Han et al 2020;Liu et al 2019a;2020a). Deep learning has drastically transformed the way in which features are extracted and then fed into a prediction model into simultaneously learning both features and a prediction model in an end-to-end fashion.…”

Section: Related Workmentioning

confidence: 99%

Subtype-aware Unsupervised Domain Adaptation for Medical Diagnosis

Liu¹,

Liu²,

Hu³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Recent advances in unsupervised domain adaptation (UDA) show that transferable prototypical learning presents a powerful means for class conditional alignment, which encourages the closeness of cross-domain class centroids. However, the cross-domain inner-class compactness and the underlying fine-grained subtype structure remained largely underexplored. In this work, we propose to adaptively carry out the fine-grained subtype-aware alignment by explicitly enforcing the class-wise separation and subtype-wise compactness with intermediate pseudo labels. Our key insight is that the unlabeled subtypes of a class can be divergent to one another with different conditional and label shifts, while inheriting the local proximity within a subtype. The cases with or without the prior information on subtype numbers are investigated to discover the underlying subtype structure in an online fashion. The proposed subtype-aware dynamic UDA achieves promising results on a medical diagnosis task.

show abstract

Wasserstein Loss based Deep Object Detection

Cited by 29 publications

References 24 publications

United We Learn Better: Harvesting Learning Improvements From Class Hierarchies Across Tasks

United We Learn Better: Harvesting Learning Improvements From Class Hierarchies Across Tasks

Embedding Semantic Hierarchy in Discrete Optimal Transport for Risk Minimization

Subtype-aware Unsupervised Domain Adaptation for Medical Diagnosis

Contact Info

Product

Resources

About