Neural Video Compression has emerged in recent years, with condition-based frameworks outperforming traditional codecs. However, most existing methods rely solely on the previous frame's features to predict temporal context, leading to two critical issues. First, the short reference window misses long-term dependencies and fine texture details. Second, propagating only feature-level information accumulates errors over frames, causing prediction inaccuracies and loss of subtle textures. To address these, we propose the Long-term Spatio-Temporal Enhanced Context (L-STEC) method. We first extend the reference chain with LSTM to capture long-term dependencies. We then incorporate warped spatial context from the pixel domain, fusing spatio-temporal information through a multi-receptive field network to better preserve reference details. Experimental results show that L-STEC significantly improves compression by enriching contextual information, achieving 37.01% bitrate savings in PSNR and 31.65% in MS-SSIM compared to DCVC-TCM, outperforming both VTM-17.0 and DCVC-FM and establishing new state-of-the-art performance.

This study addresses the challenge of domain adaptation in learned image compression, focusing on shifting the model from natural images to user-generated content (UGC) domain. We propose a novel entropy adapter framework augmented with knowledge distillation techniques to improve performance. Unlike existing adapter-based methods that primarily enhance transformation modules, we identify the mismatch between the adapter-based transformation and the fixed entropy network. To resolve this, we introduce adapters within the hypernet and entropy model. Specifically, our decoupled entropy adapter features a deeper residual structure with two independent branches, enabling a separate refinement of mean and scale components. This design improves the accuracy of probability estimation and overall compression efficiency. To further enhance the effectiveness of the adapters, we incorporate a knowledge distillation (KD) strategy with a progressive loss function. It facilitates a smooth transition from KD loss to a rate-distortion (RD) loss in the training process, effectively transferring knowledge from a directly fine-tuned model to the student model. Consequently, this strengthens the adapter's learning capability and improves compression performance. Experimental results show that the proposed method achieves a significant 11.5% bitrate savings compared to the baseline model. Additionally, it demonstrates robust adaptability across diverse network architectures.

As an incubator of next generation video coding techniques beyond versatile video coding (VVC) capability, enhanced compression model (ECM) has been initiated by the Joint Video Exploration Team (JVET). This paper presents an Inter CCP merge mode to improve the coding performance for chroma inter coding. Experimental results show that Inter CCP merge mode provides an average Bjontegaard delta rate (BD-rate) change of 0.01%/-0.69%/-0.76% and -0.03%/-2.31%/-2.38% on Y/Cb/Cr components, respectively, compared with ECM-10.0 in RA/LDB configurations under the common test condition, with a negligible running time change.

Geometric partitioning mode (GPM) splits a coding block into two partitions, which can be non-rectangular, separated by a straight splitting line. Two uni inter-predictions generated by translational motion compensation (TMC) for the two GPM partitions are blended to obtain the final prediction. With a promising coding gain, GPM has been adopted in versatile video coding (VVC). Beyond VVC, enhanced compression model (ECM) introduces several extensions on GPM, but GPM still cannot deal with affine motions well. This paper presents a method of GPM with affine prediction (GPM-affine). A GPM partition can be predicted by affine motion compensation (AMC) or TMC, indicated by a flag. A GPM partition predicted by AMC can be blended with the other GPM partition predicted by AMC, TMC, or intra-prediction. Experimental results show that GPM-affine provides an average luma BD-rate saving of 0.19% compared to ECM-10.0 in random access configurations under the common test condition, with a negligible running time change. On sequences with rich affine motions, 1% coding gain in average is observed. Currently, GPM-affine is under study in exploration experiments (EE) for ECM in JVET.