In this paper, we propose a novel feature map compression method for Video Coding for Machines (VCM). The proposed method performs a principal component analysis (PCA)-based transform on feature pyramid network (FPN) feature maps using predefined basis and mean vectors. In addition, the proposed method reduces redundancy between different resolution levels within FPN feature maps based on redundancy between FPN layers. The fixed predefined basis and mean are employed through PCA with a set of training data set. For any input videos, transform coefficients are obtained by performing transform with the fixed basis and compressed using Versatile Video Coding (VVC). Experimental results show that the proposed method achieves 89.22% and 86.57% BD-rate gain compared to the VCM feature anchor in instance segmentation, and object detection, respectively.
KEYWORDS: Video, Video compression, Machine vision, Distortion, Video coding, Signal processing, Image compression, Visual process modeling, Networks, Image processing
We previously trained the compression network via optimization of bit-rate and distortion (feature domain MSE) [1]. In this paper, we propose feature map compression method for video coding for machine (VCM) based on deep learning-based compression network that joint training for optimizing both compressed bit rate and machine vision task performance. We use bmshij2018-hyperporior model in the CompressAI [2] as the compression network, and compress the feature map which is the output of stem layer in the Faster R-CNN X101-FPN network of Detectron2 [3]. We evaluated the proposed method by evaluation framework for MPEG VCM. The proposed method shows the better results than VVC of MPEG VCM anchor.
For the higher coding performance than the previous video coding standards, high
efficiency video coding (HEVC) adopts an angular intra prediction method, which
requires heavy computational complexity due to the increased intra prediction
modes. In this paper, we propose a fast intra prediction mode decision based on
the estimation of rate distortion cost using Hadamard transform to reduce the
number of intra prediction mode and early termination whether the current coding
unit is splitted or not. The experimental results show that the proposed method
reduces the computational complexity of intra prediction in HEVC and achieves
similar coding performance to that of HEVC test mode 2.1
The previous steganographic algorithm results in the modification of the image statistic, especially histogram of
the coefficients. We propose a method to compensate for the histogram modification due to embedding. The
new algorithm estimates the modification before the embedding process has started and modifies the histogram
in advance so that the modification due to embedding is less noticeable. We have implemented our methods in
Java and performed the extensive experiments with them. The experimental results have shown that our new
method improves our previous steganographic algorithms by decreasing distortion and histogram modification.
It is well known that all information hiding methods that modify the least significant bits introduce distortions into the cover objects. Those distortions have been utilized by steganalysis algorithms to detect that the objects had been modified. It has been proposed that only coefficients whose modification does not introduce large distortions should be used for embedding. In this paper we propose an effcient algorithm for information hiding in the LSBs of JPEG coefficients. Our algorithm uses parity coding to choose the coefficients whose modifications introduce minimal additional distortion. We derive the expected value of the additional distortion as a function of the message length and the probability distribution of the JPEG quantization errors of cover images. Our experiments show close agreement between the theoretical prediction and the actual additional distortion.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.