Finite scalar quantization simplifies VQ-VAE latents by independently rounding a few dimensions to fixed levels, producing an equivalent-sized implicit codebook with competitive performance and no collapse.
Lossy Image Compression with Compressive Autoencoders
8 Pith papers cite this work. Polarity classification is still indexing.
8
Pith papers citing it
abstract
We propose a new approach to the problem of optimizing autoencoders for lossy image compression. New media formats, changing hardware technology, as well as diverse requirements and content types create a need for compression algorithms which are more flexible than existing codecs. Autoencoders have the potential to address this need, but are difficult to optimize directly due to the inherent non-differentiabilty of the compression loss. We here show that minimal changes to the loss are sufficient to train deep autoencoders competitive with JPEG 2000 and outperforming recently proposed approaches based on RNNs. Our network is furthermore computationally efficient thanks to a sub-pixel architecture, which makes it suitable for high-resolution images. This is in contrast to previous work on autoencoders for compression using coarser approximations, shallower architectures, computationally expensive methods, or focusing on small images.
representative citing papers
SAD is a new explicit differentiable image representation based on soft anisotropic additively weighted Voronoi partitions that achieves higher PSNR and 4-19x faster training than Image-GS and Instant-NGP at matched bitrate.
Few-step generative models can be reformulated as lossy codecs in the reverse channel coding framework without retraining, yielding faster encoding/decoding on low-resolution image benchmarks.
LANCE extends OIC frameworks with a spatial hyperprior and predictive coding scheme, reporting BD-rate gains of 1.4-3% over Cool-Chic 4.0 on Kodak and CLIC.
A deep convolutional autoencoder compression framework jointly optimized with face recognition achieves higher verification accuracy on LFW images than JPEG2000 or JPEG.
Derives optimality constraints for nonnegative joint dictionary learning that explain observed SAE behaviors such as feature splitting, absorption, and dense antipodal features.
Cool-chic 5.0 delivers 11% lower rate than H.266/VVC and matches modern autoencoders like MLIC++ with 250 times lower decoding complexity through an updated decoder architecture and faster optimization for overfitted codecs.
DinoLink uses saliency-aware token pruning plus residual vector quantization to cut V2X bitrate by 139x while reporting 32.8% mAP on nuScenes.
citing papers explorer
-
Finite Scalar Quantization: VQ-VAE Made Simple
Finite scalar quantization simplifies VQ-VAE latents by independently rounding a few dimensions to fixed levels, producing an equivalent-sized implicit codebook with competitive performance and no collapse.
-
Soft Anisotropic Diagrams for Differentiable Image Representation
SAD is a new explicit differentiable image representation based on soft anisotropic additively weighted Voronoi partitions that achieves higher PSNR and 4-19x faster training than Image-GS and Instant-NGP at matched bitrate.
-
Few-step Generative Models as Lossy Compression
Few-step generative models can be reformulated as lossy codecs in the reverse channel coding framework without retraining, yielding faster encoding/decoding on low-resolution image benchmarks.
-
LANCE: Locally Adaptive Neural Context Estimation for Overfitted Image Compression
LANCE extends OIC frameworks with a spatial hyperprior and predictive coding scheme, reporting BD-rate gains of 1.4-3% over Cool-Chic 4.0 on Kodak and CLIC.
-
A Deep Image Compression Framework for Face Recognition
A deep convolutional autoencoder compression framework jointly optimized with face recognition achieves higher verification accuracy on LFW images than JPEG2000 or JPEG.
-
How Optimality Structures Sparse Dictionaries: A Theory for Understanding SAE Representations
Derives optimality constraints for nonnegative joint dictionary learning that explain observed SAE behaviors such as feature splitting, absorption, and dense antipodal features.
-
Cool-chic 5.0: Faster Encoding and Inter-Feature Entropy Modeling for Overfitted Image Compression
Cool-chic 5.0 delivers 11% lower rate than H.266/VVC and matches modern autoencoders like MLIC++ with 250 times lower decoding complexity through an updated decoder architecture and faster optimization for overfitted codecs.
-
DinoLink: A Token-Centric Representation Compression Framework for Bandwidth-Constrained Collaborative V2X Perception
DinoLink uses saliency-aware token pruning plus residual vector quantization to cut V2X bitrate by 139x while reporting 32.8% mAP on nuScenes.