Efficient person search via learning-to-normalize deep representation

Lv, Ning; Xiang, Xuezhi; Wang, Xinyao; Abdein, Rokia

doi:10.1016/j.neucom.2022.01.028

Cited by 2 publications

(3 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The network can optimize the expected log-likelihood by calculating the probability of labeled and unlabeled identities’ online instance matching loss, which can take into account the unlabeled identities [ 41 ] in the dataset and achieve better results. Liu et al constructed a natural detection network (NSN) [ 42 ] that can preserve spatial information from spatio-temporal sequences based on a convolutional long-short network [ 43 ] (Conv-LSTM), which can integrate the pedestrian features to be queried into the network memory and target the location of the target task by recursively narrowing the region to be retrieved. This is a novel approach that eliminates the step of pedestrian detection by combining contextual information, and it can eliminate the error accumulation effect of pedestrian detection.…”

Section: Related Workmentioning

confidence: 99%

“…Chen et al optimized the feature differentiation method [ 35 ] by transforming features to polar coordinates so that the background has as little effect on pedestrian features as possible and achieved good results by optimizing feature distances [ 41 ] in different spaces. Han et al designed an ROI [ 46 ] (region of interest) transform layer, which effectively implements the transformation of the detection frame [ 42 ] from the original image to achieve end-to-end training. Chen et al proposed Hierarchical Online Instance Matching (HOIM) [ 40 ] loss, which pairs detection and re-ID to guide feature learning in their network.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Inception Convolution and Feature Fusion for Person Search

Ouyang

Zeng

Leng

2023

Sensors

View full text Add to dashboard Cite

With the rapid advancement of deep learning theory and hardware device computing capacity, computer vision tasks, such as object detection and instance segmentation, have entered a revolutionary phase in recent years. As a result, extremely challenging integrated tasks, such as person search, might develop quickly. The majority of efficient network frameworks, such as Seq-Net, are based on Faster R-CNN. However, because of the parallel structure of Faster R-CNN, the performance of re-ID can be significantly impacted by the single-layer, low resolution, and occasionally overlooked check feature diagrams retrieved during pedestrian detection. To address these issues, this paper proposed a person search methodology based on an inception convolution and feature fusion module (IC-FFM) using Seq-Net (Sequential End-to-end Network) as the benchmark. First, we replaced the general convolution in ResNet-50 with the new inception convolution module (ICM), allowing the convolution operation to effectively and dynamically distribute various channels. Then, to improve the accuracy of information extraction, the feature fusion module (FFM) was created to combine multi-level information using various levels of convolution. Finally, Bounding Box regression was created using convolution and the double-head module (DHM), which considerably enhanced the accuracy of pedestrian retrieval by combining global and fine-grained information. Experiments on CHUK-SYSU and PRW datasets showed that our method has higher accuracy than Seq-Net. In addition, our method is simpler and can be easily integrated into existing two-stage frameworks.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Inception Convolution and Feature Fusion for Person Search

Ouyang

Zeng

Leng

2023

Sensors

View full text Add to dashboard Cite

show abstract

“…They play a central role in many downstream tasks about instance-level understanding, e.g. person search [1][2][3], visual reasoning [4,5], and human pose estimation [6][7][8]. Over the past few years, researches on object detection and instance segmentation have witnessed remarkable progresses, yielding good returns in flexible frameworks such as Faster region based convolutional neural networks (R-CNN) [9] and Mask R-CNN [10], and excellent performance on public benchmarks such as PASCAL-VOC [11] and MS-COCO [12].…”

Section: Introductionmentioning

confidence: 99%

DIM: long-tailed object detection and instance segmentation via dynamic instance memory

Chen,

Jin,

Zhang

et al. 2023

Mach. Learn.: Sci. Technol.

View full text Add to dashboard Cite

Object detection and instance segmentation have been successful on benchmarks with relatively balanced category distribution (e.g., MSCOCO). However, state-of-the-art object detection and segmentation methods still struggle to generalize on long-tailed datasets (e.g., LVIS), where a few classes (head classes) dominate the instance samples, while most classes (tailed classes) have only a few samples. To address this challenge, we propose a plug-and-play module within the Mask R-CNN framework called Dynamic Instance Memory (DIM). Specifically, we augment Mask R-CNN with an auxiliary branch for training. It maintains a dynamic memory bank storing an instance-level prototype representation for each category, and shares the classifier with the existing instance branch. With a simple metric loss, the representations in DIM can be dynamically updated by the instance proposals in the mini-batch during training. Our DIM introduces a bias toward tailed classes to the classifier learning along with a class frequency reversed sampler, which learns generalizable representations from the original data distribution, complementing the existing instance branch. Comprehensive experiments on LVIS demonstrate the effectiveness of DIM, as well as the significant advantages of DIM over the baseline Mask R-CNN.

show abstract

Efficient person search via learning-to-normalize deep representation

Cited by 2 publications

References 12 publications

Inception Convolution and Feature Fusion for Person Search

Inception Convolution and Feature Fusion for Person Search

DIM: long-tailed object detection and instance segmentation via dynamic instance memory

Contact Info

Product

Resources

About