Body Part Based Re-Identification from an Egocentric Perspective

Fergnani, Federica; Alletto, Stefano; Serra, Giuseppe; Mira, Joaquim De; Cucchiara, Rita

doi:10.1109/cvprw.2016.51

Cited by 8 publications

(22 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [2], facial features were extracted and similarities were computed between a pair of cameras to yield optimal and consistent re-ID results. People are identified by dividing their images into meaningful body parts in [3]. Furthermore, the contribution of different body parts is calculated by taking into account human gaze information.…”

Section: Egocentric Person Re-idmentioning

confidence: 99%

“…Our work utilizes an egocentric dataset shot through a single camera in [3]. Rather than depending on different body parts, we take into account the complete image of a person while re-identifying him.…”

Section: Egocentric Person Re-idmentioning

confidence: 99%

“…The first step in our model is the generation of images which has combined features of fixed camera dataset (Market-1501 [20]) and first-person vision dataset (EGO Re-ID [3]). Our model makes use of a VGG-16 [12] based Neural Style Transfer (NST) model to produce high-quality images in a manner that helps in bringing the qualities of Market-1501 [20] into EGO Re-ID [3] while preserving the content of EGO Re-ID [3]. This is achieved by formulating image generation problem as an optimization one that minimizes the following losses: (1) Content Loss, (2) Style Loss and (3) Total Loss.…”

Section: Approachmentioning

confidence: 99%

“…The existing literature for first-person re-ID i.e. body parts based [3] and multiple views based [2] methods also suffer from dataset bias.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Domain Adaptive Egocentric Person Re-identification

Choudhary,

Mishra,

Karmakar

2021

Preprint

View full text Add to dashboard Cite

Person re-identification (re-ID) in first-person (egocentric) vision is a fairly new and unexplored problem. With the increase of wearable video recording devices, egocentric data becomes readily available, and person re-identification has the potential to benefit greatly from this. However, there is a significant lack of large scale structured egocentric datasets for person re-identification, due to the poor video quality and lack of individuals in most of the recorded content. Although a lot of research has been done in person re-identification based on fixed surveillance cameras, these do not directly benefit egocentric re-ID. Machine learning models trained on the publicly available large scale re-ID datasets cannot be applied to egocentric re-ID due to the dataset bias problem. The proposed algorithm makes use of neural style transfer (NST) that incorporates a variant of Convolutional Neural Network (CNN) to utilize the benefits of both fixed camera vision and firstperson vision. NST generates images having features from both egocentric datasets and fixed camera datasets, that are fed through a VGG-16 network trained on a fixed-camera dataset for feature extraction. These extracted features are then used to re-identify individuals. The fixed camera dataset Market-1501 [20] and the first-person dataset EGO Re-ID [3] are applied for this work and the results are on par with the present re-identification models in the egocentric domain.

show abstract

Section: Egocentric Person Re-idmentioning

confidence: 99%

Section: Egocentric Person Re-idmentioning

confidence: 99%

Section: Approachmentioning

confidence: 99%

“…The existing literature for first-person re-ID i.e. body parts based [3] and multiple views based [2] methods also suffer from dataset bias.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Domain Adaptive Egocentric Person Re-identification

Choudhary,

Mishra,

Karmakar

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…D Kouno et al [22] proposed utilizing depth information for the problem based on an image from an overhead camera by decreasing the influence of occluded images. Fergnani et al [13] exploited the unartificial proportions of person body composed by a division of body parts, which only relies on a real-time estimation of facial symbol and not requires complex transmutable part models.…”

Section: Introductionmentioning

confidence: 99%

Manifold ranking weighted local maximal occurrence descriptor for person re-identification

Wang

Zhang

et al. 2017

2017 IEEE 15th International Conference on Software Engineering Research, Management and Applications (SERA)

View full text Add to dashboard Cite

Person re-identification is an important task of matching pedestrians across non-overlapping camera views. In this paper, we exploit a weighted feature descriptor for person re-identification. We firstly compute the weights on the superpixel level via graph-based manifold ranking algorithm, then integrate the computed weights into a patch-based feature descriptor, named local maximal occurrence. Finally, the weighted descriptors are fed into a top-push distance learning to mitigate the cross-view gaps. We evaluate the proposed method on three benchmark datasets iLIDS-VID, PRID 450S and VIPeR. The promising experimental results demonstrate the effectiveness of the proposed method comparing with the state-of-the-arts.

show abstract