IJB–S: IARPA Janus Surveillance Video Benchmark

Kalka, Nathan D.; Maze, Brianna; Duncan, James A.; OrConnor, Kevin; Elliott, Stephen J.; Hebert, Kaleb; Bryan, Julia; Jain, Anil K.

doi:10.1109/btas.2018.8698584

Cited by 52 publications

(65 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Surveillance-to-Single Surveillance-to-Booking Surveillance-to-Surveillance Rank-1 Rank-5 Rank-10 1% 10% Rank- Table 5: Performance comparison on three protocols of IJB-S. The performance is reported in terms of rank retrieval (closed-set) and TPIR@FPIR (open-set) instead of the media-normalized version [13]. The numbers "1%" and "10%" in the second row refer to the FPIR.…”

Section: Training Datamentioning

confidence: 99%

Probabilistic Face Embeddings

Shi

Jain

2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

272

183

View full text Add to dashboard Cite

Embedding methods have achieved success in face recognition by comparing facial features in a latent semantic space. However, in a fully unconstrained face setting, the facial features learned by the embedding model could be ambiguous or may not even be present in the input face, leading to noisy representations. We propose Probabilistic Face Embeddings (PFEs), which represent each face image as a Gaussian distribution in the latent space. The mean of the distribution estimates the most likely feature values while the variance shows the uncertainty in the feature values. Probabilistic solutions can then be naturally derived for matching and fusing PFEs using the uncertainty information. Empirical evaluation on different baseline models, training datasets and benchmarks show that the proposed method can improve the face recognition performance of deterministic embeddings by converting them into PFEs. The uncertainties estimated by PFEs also serve as good indicators of the potential matching accuracy, which are important for a risk-controlled recognition system.

show abstract

Section: Training Datamentioning

confidence: 99%

Probabilistic Face Embeddings

Shi

Jain

2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

272

183

View full text Add to dashboard Cite

show abstract

“…The baseline detector is trained on the WIDER Train split, which has 12,880 images and 159,424 annotated faces. The target domain consists of 179 surveillance videos from CS6, which is a subset of the IJB-S benchmark [28]. CS6 provides a considerable shift from WIDER, with faces being mostly low-resolution and often occluded, and the imagery being of low picture quality, suffering from camera shake and motion blurs.…”

Section: Datasetsmentioning

confidence: 99%

“…Due to the data-hungry nature of supervised training of deep networks, it would require a lot of label- Figure 1: Unsupervised cross-domain object detection. Top: adapting a face detector trained on labeled highquality web images from WIDER-Face [64] to unlabeled CS6/IJB-S [28] video frames. Bottom: adapting a pedestrian detector trained on labeled images from the (clear, daytime) split of the BDD-100k dataset [65] to unlabeled videos from all the other conditions (e.g.…”

Section: Introductionmentioning

confidence: 99%

Automatic Adaptation of Object Detectors to New Domains Using Self-Training

RoyChowdhury

Chakrabarty

Singh

et al. 2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

146

View full text Add to dashboard Cite

This work addresses the unsupervised adaptation of an existing object detector to a new target domain. We assume that a large number of unlabeled videos from this domain are readily available. We automatically obtain labels on the target data by using high-confidence detections from the existing detector, augmented with hard (misclassified) examples acquired by exploiting temporal cues using a tracker. These automatically-obtained labels are then used for re-training the original model. A modified knowledge distillation loss is proposed, and we investigate several ways of assigning soft-labels to the training examples from the target domain. Our approach is empirically evaluated on challenging face and pedestrian detection tasks: a face detector trained on WIDER-Face, which consists of highquality images crawled from the web, is adapted to a largescale surveillance data set; a pedestrian detector trained on clear, daytime images from the BDD-100K driving data set is adapted to all other scenarios such as rainy, foggy, nighttime. Our results demonstrate the usefulness of incorporating hard examples obtained from tracking, the advantage of using soft-labels via distillation loss versus hard-labels, and show promising performance as a simple method for unsupervised domain adaptation of object detectors, with minimal dependence on hyper-parameters. Code and models are available at

show abstract

“…Our approach can be directly applied at inference time, or plugged onto an end-to-end network architecture for supervised and semi-supervised training. The proposed method is evaluated on two challenging datasets, the Cast Search in Movies (CSM) dataset [11] and the IARPA Janus Surveillance Video Benchmark (IJB-S) dataset [12] with superior performance compared to existing methods.…”

Section: Still Face Galleriesmentioning

confidence: 99%

“…Probes are remotely captured surveillance videos from which all the tracklets are required. We report the per tracklet average top-K identification accuracy and the End-to-End Retrieval Rate (EERR) metric proposed in [12] for performance evaluation. Please refer [12] for more details.…”

Section: Csmmentioning

confidence: 99%

Uncertainty Modeling of Contextual-Connections Between Tracklets for Unconstrained Video-Based Face Recognition

Zheng

Rong

Chen

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

View full text Add to dashboard Cite

Unconstrained video-based face recognition is a challenging problem due to significant within-video variations caused by pose, occlusion and blur. To tackle this problem, an effective idea is to propagate the identity from highquality faces to low-quality ones through contextual connections, which are constructed based on context such as body appearance. However, previous methods have often propagated erroneous information due to lack of uncertainty modeling of the noisy contextual connections. In this paper, we propose the Uncertainty-Gated Graph (UGG), which conducts graph-based identity propagation between tracklets, which are represented by nodes in a graph. UGG explicitly models the uncertainty of the contextual connections by adaptively updating the weights of the edge gates according to the identity distributions of the nodes during inference. UGG is a generic graphical model that can be applied at only inference time or with end-to-end training. We demonstrate the effectiveness of UGG with state-of-theart results in the recently released challenging Cast Search in Movies and IARPA Janus Surveillance Video Benchmark dataset.

show abstract

IJB–S: IARPA Janus Surveillance Video Benchmark

Cited by 52 publications

References 19 publications

Probabilistic Face Embeddings

Probabilistic Face Embeddings

Automatic Adaptation of Object Detectors to New Domains Using Self-Training

Uncertainty Modeling of Contextual-Connections Between Tracklets for Unconstrained Video-Based Face Recognition

Contact Info

Product

Resources

About