Fine-Grained Multi-Class Object Counting

Go, Hyojun; Byun, Junyoung; Park, Byeongjun; Choi, Moon; Yoo, Seunghwa; Kim, Changick

doi:10.1109/icip42928.2021.9506384

Cited by 16 publications

(18 citation statements)

References 12 publications

(17 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Wan et al [15] extend this task to the multi-class setting by adding fine-grained attributes to existing human counting datasets. Go et al [6] introduce KR-GRUIDAE, a fine-grained counting dataset consisting of 5 bird classes. In comparison to existing datasets for fine-grained counting, Seal Watch contains more classes, images, object instances, and annotations.…”

Section: Related Workmentioning

confidence: 99%

“…In this work we study the use of crowd-sourced image annotations to train computer vision algorithms on the challenging task of fine-grained counting. Recently introduced in [6,15], fine-grained counting extends crowd countingestimating the number of individuals in a densely crowded scene, typically used for counting human crowds-to a fine- grained multi-class scenario. While previous work has proposed methods for utilizing crowd-sourced annotations for image classification [10] and single-class counting [3], we are the first to study their use for training algorithms for fine-grained counting.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Fine-Grained Counting with Crowd-Sourced Supervision

Kay¹,

Foley²,

Hart³

2022

Preprint

View full text Add to dashboard Cite

Crowd-sourcing is an increasingly popular tool for image analysis in animal ecology. Computer vision methods that can utilize crowd-sourced annotations can help scale up analysis further. In this work we study the potential to do so on the challenging task of fine-grained counting. As opposed to the standard crowd counting task, fine-grained counting also involves classifying attributes of individuals in dense crowds. We introduce a new dataset from animal ecology to enable this study that contains 1.7M crowdsourced annotations of 8 fine-grained classes. It is the largest available dataset for fine-grained counting and the first to enable the study of the task with crowd-sourced annotations. We introduce methods for generating aggregate "ground truths" from the collected annotations, as well as a counting method that can utilize the aggregate information. Our method improves results by 8% over a comparable baseline, indicating the potential for algorithms to learn fine-grained counting using crowd-sourced supervision.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Fine-Grained Counting with Crowd-Sourced Supervision

Kay¹,

Foley²,

Hart³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Current multi-class systems operate on known classes. They achieve this using either detection directly on the image [6,20] or on a regressed density map [18,49].…”

Section: Related Workmentioning

confidence: 99%

“…Previous methods, whether detection-based, regression-based or classification-based, generally focus on enumerating the instances of a single or small set of known classes, such as people [4,7,32,33,45,47], vehicles [1,32,35], animals [2,18], or cells [1,32,48]. This requires an individually trained network for each type of object with limited to no capacity to adapt to previously unseen classes.…”

Section: Introductionmentioning

confidence: 99%

Learning to Count Anything: Reference-less Class-agnostic Counting with Weak Supervision

Hobley¹,

Prisacariu²

2022

Preprint

View full text Add to dashboard Cite

Object counting is a seemingly simple task with diverse real-world applications. Most counting methods focus on counting instances of specific, known classes. While there are class-agnostic counting methods that can generalise to unseen classes, these methods require reference images to define the type of object to be counted, as well as instance annotations during training. We identify that counting is, at its core, a repetition-recognition task and show that a general feature space, with global context, is sufficient to enumerate instances in an image without a prior on the object type present. Specifically, we demonstrate that self-supervised vision transformer features combined with a lightweight count regression head achieve competitive results when compared to other class-agnostic counting tasks without the need for point-level supervision or reference images. Our method thus facilitates counting on a constantly changing set composition. To the best of our knowledge, we are both the first reference-less class-agnostic counting method as well as the first weakly-supervised class-agnostic counting method.

show abstract

“…While these cameras automatically collect massive data, identifying species by humans is timeconsuming and labor-intensive, limiting research productivity. Therefore, deep neural networks [2,3] have recently received attention for their ability to automate the identification process, making camera-trap studies scalable [4,5]. Nevertheless, neural networks tend to be biased towards the species that frequently appear, limiting studies that require diverse animal species, specifically on endangered species.…”

Section: Introductionmentioning

confidence: 99%

Balancing Domain Experts for Long-Tailed Camera-Trap Recognition

Park¹,

Kim²,

Cho³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Label distributions in camera-trap images are highly imbalanced and long-tailed, resulting in neural networks tending to be biased towards head-classes that appear frequently. Although long-tail learning has been extremely explored to address data imbalances, few studies have been conducted to consider camera-trap characteristics, such as multi-domain and multi-frame setup. Here, we propose a unified framework and introduce two datasets for long-tailed camera-trap recognition. We first design domain experts, where each expert learns to balance imperfect decision boundaries caused by data imbalances and complement each other to generate domain-balanced decision boundaries. Also, we propose a flow consistency loss to focus on moving objects, expecting class activation maps of multi-frame matches the flow with optical flow maps for input images. Moreover, two long-tailed camera-trap datasets, WCS-LT and DMZ-LT, are introduced to validate our methods. Experimental results show the effectiveness of our framework, and proposed methods outperform previous methods on recessive domain samples.

show abstract

Fine-Grained Multi-Class Object Counting

Cited by 16 publications

References 12 publications

Fine-Grained Counting with Crowd-Sourced Supervision

Fine-Grained Counting with Crowd-Sourced Supervision

Learning to Count Anything: Reference-less Class-agnostic Counting with Weak Supervision

Balancing Domain Experts for Long-Tailed Camera-Trap Recognition

Contact Info

Product

Resources

About