Quantized Decoder in Learned Image Compression for Deterministic Reconstruction
Abstract: Learned image compression has a problem of non-bit-exact reconstruction due to different calculations of floating point arithmetic on different devices. This paper shows a method to achieve a deterministic reconstructed image by quantizing only the decoder of the learned image compression model. From the implementation perspective of an image codec, it is beneficial to have the results reproducible when decoded on different devices. In this paper, we study quantization of weights and activations without overflow of accumulator in all decoder subnetworks. We show that the results are bit-exact at the output, and the resulting BD-rate loss of quantization of decoder is 0.5 % in the case of 16-bit weights and 16-bit activations, and 7.9 % in the case of 8-bit weights and 16-bit activations.
- Gregory K Wallace, “The JPEG still picture compression standard,” Communications of the ACM, vol. 34, no. 4, pp. 30–44, 1991.
- “The JPEG 2000 still image compression standard,” IEEE Signal Processing Magazine, vol. 18, no. 5, pp. 36–58, 2001.
- “Overview of the high efficiency video coding (HEVC) standard,” IEEE Trans. Circuits Syst. Video Technol., vol. 22, no. 12, pp. 1649–1668, 2012.
- “Overview of the versatile video coding (VVC) standard and its applications,” IEEE Trans. Circuits Syst. Video Technol., vol. 31, no. 10, pp. 3736–3764, 2021.
- “End-to-end optimized image compression,” in 5th Int. Conf. on Learning Representations, ICLR 2017, 2017.
- “Variational image compression with a scale hyperprior,” in Proc. Int. Conf. on Learning Representations, 2018.
- “Joint autoregressive and hierarchical priors for learned image compression,” in Neural Information Processing Systems, 2018.
- “Learned image compression with discretized Gaussian mixture likelihoods and attention modules,” in Proceedings of the IEEE/CVF Conf. on Comput. Vis. Pattern Recog., 2020, pp. 7939–7948.
- “Learned image compression with mixed transformer-CNN architectures,” in Proceedings of the IEEE/CVF Conf. on Comput. Vis. Pattern Recog., 2023, pp. 14388–14397.
- “Efficient Neural Image Decoding via Fixed-Point Inference,” IEEE Trans. Circuits Syst. Video Technol., vol. 31, pp. 3618–3630, Sept. 2021.
- “Q-LIC: Quantizing Learned Image Compression with Channel Splitting,” IEEE Trans. Circuits Syst. Video Technol., pp. 1–1, 2022.
- “MobileCodec: Neural inter-frame video compression on mobile devices,” in Proceedings of the 13th ACM Multimedia Systems Conference, New York, NY, USA, Aug. 2022, MMSys ’22, pp. 324–330, Association for Computing Machinery.
- “End-To-End Learned Image Compression With Fixed Point Weight Quantization,” in 2020 IEEE Int. Conf. on Image Processing (ICIP), Abu Dhabi, United Arab Emirates, Oct. 2020, pp. 3359–3363.
- “Learned image compression with fixed-point arithmetic,” in 2021 Picture Coding Symposium (PCS), Bristol, United Kingdom, June 2021, pp. 1–5.
- “Integer networks for data compression with latent-variable models,” in Proc. Int. Conf. on Learning Representations, 2018.
- “Device interoperability for learned image compression with weights and activations quantization,” in 2022 Picture Coding Symposium (PCS), 2022, pp. 151–155.
- “JPEG AI verification model,” [Online]. Available ” https://jpeg-git.lx.it.pt/jpegai/vm”.
- “Asymmetric gained deep image compression with continuous rate adaptation,” in Proceedings of the IEEE/CVF Conf. on Comput. Vis. Pattern Recog., 2021, pp. 10532–10541.
- Gisle Bjøntegaard, “Calculation of average PSNR differences between RD-curves,” VCEG-M33, 2001.
- “JPEG AI dataset,” 2022, [Online]. Available ” https://jpeg.org/jpegai/dataset.html”.
- “JPEG AI common training and test conditions,” Tech. Rep., ISO/IEC JTC 1/SC 29/WG 1 (& ITU-T SG16), 2023, [Online]. Available ”https://ds.jpeg.org/documents/jpegai/wg1n100421-098-ICQ-JPEG_AI_Common_Training_%26_Test_Conditions.pdf”.
- “NVIDIA A100 tensor core GPU architecture,” [Online]. Available ”https://images.nvidia.com/aem-dam/en-zz/Solutions/data-center/nvidia-ampere-architecture-whitepaper.pdf”.
- “Snapdragon neural processing engine SDK reference guide,” [Online]. Available ”https://developer.qualcomm.com/sites/default/files/docs/snpe/quantized_models.html”.
Paper Prompts
Sign up for free to create and run prompts on this paper using GPT-5.
Top Community Prompts
Collections
Sign up for free to add this paper to one or more collections.