Distilling Global and Local Logits with Densely Connected Relations

Kim, Youmin; Park, Jin-Bae; Jang, YounHo; Ali, Muhammad; Oh, Tae-Hyun; Bae, Sung‐Ho

doi:10.1109/iccv48922.2021.00623

Cited by 22 publications

(2 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We vary the distortion rate by using different bit-depth levels = [8,4,2] for quantization. The result in the feature domain reveals no significant loss in mAP performance, but the image domain indicates a considerable drop in mAP with 2-bit quantization, as shown in Figure 7.…”

Section: F Feature Robustnessmentioning

confidence: 99%

“…With the recent development of artificial intelligence research, deep neural networks (DNN) outperform the conventional shallow models on various vision tasks such as image classification, object detection, segmentation, and tracking [2]- [4], [6]- [8]. Likewise, for video/image compression purpose, DNNs have been utilized and shown promising performance.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Super-Resolution-Based Feature Map Compression for Machine-Oriented Video Coding

et al. 2023

Self Cite

View full text Add to dashboard Cite

Recently, video and image compression methods using neural networks have received much attention. In MPEG standardization, Video Coding for Machine (VCM) is a newly arising topic which attempts to compress features/images for the purpose of machine vision tasks. Especially, compressing features has advantages in terms of privacy protection and computation off-loading. In this paper, we propose an effective feature compression method equipped with a super-resolution (SR) module for features. Our main motivation comes from the observation that features are somewhat robust to spatial distortions (e.g., AWGN, blur, quantization distortions, coding artifacts), which leads us to integrating an SR module into the compression framework. We also further explore the best training strategy of the proposed method, i.e., finding the best combination of various losses and proper input feature shapes. Our comprehensive experiments show that the proposed method outperforms the baseline in the original VCM anchor scenario on various QP values with Versatile Video Coding (VVC). Specifically, the proposed framework achieved up to 50% BD-rate reduction compared to the conventional P-layer feature map compression method for the object detection task on the OpenImage dataset.INDEX TERMS Versatile video codec, video coding for machine, feature compression, deep neural network, super resolution

show abstract

Section: F Feature Robustnessmentioning

confidence: 99%