• Corpus ID: 231741384

Latent-Space Inpainting for Packet Loss Concealment in Collaborative Object Detection

@article{Bajic2021LatentSpaceIF,
  title={Latent-Space Inpainting for Packet Loss Concealment in Collaborative Object Detection},
  author={Ivan V. Baji'c},
  journal={ArXiv},
  year={2021},
  volume={abs/2102.00142}
}
Edge devices, such as cameras and mobile units, are increasingly capable of performing sophisticated computation in addition to their traditional roles in sensing and communicating signals. The focus of this paper is on collaborative object detection, where deep features computed on the edge device from input images are transmitted to the cloud for further processing. We consider the impact of packet loss on the transmitted features and examine several ways for recovering the missing data. In… 

Figures and Tables from this paper

References

SHOWING 1-10 OF 29 REFERENCES
Deep Feature Compression for Collaborative Object Detection
  • Hyomin Choi, I. Bajić
  • Computer Science
    2018 25th IEEE International Conference on Image Processing (ICIP)
  • 2018
TLDR
This work focuses on collaborative object detection and study the impact of both near-lossless and lossy compression of feature data on its accuracy, and proposes a strategy for improving the accuracy under lossy feature compression.
Near-Lossless Deep Feature Compression for Collaborative Intelligence
  • Hyomin Choi, I. Bajić
  • Engineering, Computer Science
    2018 IEEE 20th International Workshop on Multimedia Signal Processing (MMSP)
  • 2018
TLDR
This work examines the differences between the deep feature data and natural image data, and proposes a simple and effective near-lossless deep feature compressor that achieves up to 5% bit rate reduction compared to HEVC-Intra and even more against other popular image codecs.
Latent Space Motion Analysis for Collaborative Intelligence
  • Mateen Ulhaq, I. Bajić
  • Computer Science
    ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
  • 2021
TLDR
By analyzing the effect of common DNN operations on optical flow, it is shown that the motion present in each channel of a feature tensor is approximately equal to the scaled version of the input motion.
Video Coding for Machines: A Paradigm of Collaborative Compression and Intelligent Analytics
TLDR
The definition, formulation, and paradigm of VCM are given, and the preliminary results have demonstrated the performance and efficiency gains, and state-of-the-art techniques in video compression and feature compression are systematically reviewed from the unique perspective of MPEG standardization.
Toward Intelligent Sensing: Intermediate Deep Feature Compression
TLDR
This paper proposes to compactly represent and convey the intermediate-layer deep learning features with high generalization capability, to facilitate the collaborating approach between front and cloud ends and makes the standardization of deep feature coding more feasible and promising.
Image inpainting
TLDR
A novel algorithm for digital inpainting of still images that attempts to replicate the basic techniques used by professional restorators, and does not require the user to specify where the novel information comes from.
An Image Inpainting Technique Based on the Fast Marching Method
  • A. Telea
  • Computer Science
    J. Graphics, GPU, & Game Tools
  • 2004
TLDR
A new algorithm for digital inpainting based on the fast marching method for level set applications is presented, very simple to implement, fast, and produces nearly identical results to more complex, and usually slower, known methods.
PDE-Based Image and Surface Inpainting
TLDR
This chapter is intended to present an overview of PDE based image inpainting algorithms, with emphasis in models developed by the authors, based on the propagation of information along the image isophotes and on the minimization of an energy functional which follows a relaxation of the Elastica model.
Multi-Task Learning with Compressible Features for Collaborative Intelligence
TLDR
This work introduces a new loss function that encourages feature compressibility while improving system performance on multiple tasks and shows that with the compression-friendly loss, one can achieve around 20% bitrate reduction without sacrificing the performance on several vision-related tasks.
Lightweight Compression Of Neural Network Feature Tensors For Collaborative Intelligence
TLDR
A novel lightweight compression technique designed specifically to code the activations of a split DNN layer, while having a low complexity suitable for edge devices and not requiring any retraining, is presented.
...
1
2
3
...