This paper presents a review of the LoViF 2026 Challenge on Weather Removal in Videos. The challenge encourages the development of methods for restoring clean videos from inputs degraded by adverse weather conditions such as rain and snow, with an emphasis on achieving visually plausible and temporally consistent results while preserving scene structure and motion dynamics. To support this task, we introduce a new short-form WRV dataset tailored for video weather removal. It consists of 18 videos 1,216 synthesized frames paired with 1,216 real-world ground-truth frames at a resolution of 832 x 480, and is split into training, validation, and test sets with a ratio of 1:1:1. The goal of this challenge is to advance robust and realistic video restoration under real-world weather conditions, with evaluation protocols that jointly consider fidelity and perceptual quality. The challenge attracted 37 participants and received 5 valid final submissions with corresponding fact sheets, contributing to progress in weather removal for videos. The project is publicly available at https://www.codabench.org/competitions/13462/.
Figures & tables
Rank
Team
Final Score ↑
PSNR ↑
SSIM ↑
LPIPS ↓
Warp Error ↓
1
RedMediaTech
16.21
14.751
0.554
0.624
0.032
2
tremendous
15.54
14.720
0.533
0.688
0.036
3
Guangong Perception
15.11
14.206
0.531
0.613
0.045
4
quadrillion
14.68
13.775
0.524
0.712
0.026
5
CUIT_Team
14.54
13.539
0.527
0.673
0.030
Table 1: Results of the LoViF 2026 Challenge on Weather Removal in Videos. Teams are ranked by Final Score. Best results in each metric are highlighted in bold . ↑ indicates higher is better, ↓ indicates lower is better.
Figure 1: Visual comparison on a rainy scene. From left to right, top to bottom: Ground Truth, RedMediaTech, tremendous, Guangong Perception, quadrillion, and CUIT_Team.
Figure 2: Visual comparison on a snowy scene. From left to right, top to bottom: Ground Truth, RedMediaTech, tremendous, Guangong Perception, quadrillion, and CUIT_Team.
Figure 3: Overview of the RedMediaTech method. Stage 1: the DiT backbone is pre-trained on FoundIR with MSE and LPIPS losses. Stage 2: the DiT is fine-tuned on the competition data using LoRA with MSE loss. The VAE remains frozen throughout.
Figure 4: Architecture of the TDyConv module proposed by team tremendous. A Transformer encoder generates kernel attention maps that dynamically modulate depthwise convolution weights, enabling content-adaptive processing of heterogeneous weather degradations.
Figure 5: Overall framework of VP-AdaIR proposed by team Guangong Perception. The method takes a degraded video clip as input, extracts a three-channel temporal prior map (background, artifact, and temporal stability cues), and injects it into the AdaIR backbone through prior-guided fusion.
Figure 6: Overall architecture of EF 3 Net proposed by team quadrillion. A Siamese Historyless Encoder independently processes previous and current frames. A Causal History Model (CHM) dynamically fuses temporal memory via Key/Value caching at the bottleneck.
Figure 7: Overall architecture of VD-Diff proposed by CUIT_Team. The framework consists of a conditional diffusion model for implicit prior generation and a wavelet-decomposition-based restoration backbone with Adaptive Window Self-Attention (AWSA), Adaptive Channel Self-Attention (ACSA), and bidirectional temporal propagation.
Adverse weather removal (AWR) in real-world images remains challenging due to heterogeneous and unseen degradations, while distortion-driven training often yields overly smooth results. We propose PVRF, a unified framework that integrates zero-shot soft weather perceptions with velocity-constrained rectified-flow refinement. PVRF introduces an AWR-specific question answering module (AWR-QA) that uses frozen vision--language models (VLMs) to estimate soft probabilities of weather types and low-level attribute scores. These perceptions condition restoration networks via attribute-modulated normalization (AMN) and weather-weighted adapters (WWA), producing an anchor estimate for refinement. We then learn a terminal-consistent residual rectified flow with perception-adaptive source perturbation and a terminal-consistent velocity parameterization to stabilize learning near the terminal regime. Extensive experiments show that PVRF improves both fidelity and perceptual quality over state-of-the-art baselines, with strong cross-dataset generalization on single and combined degradations. Code will be released at https://github.com/dongw22/PVRF.
Wei Dong, Han Zhou, Terry Ji +6
McMaster University · Shanghai Jiao Tong University
Existing video deraining methods typically exploit neighboring frames through either explicit alignment or implicit spatiotemporal aggregation. Explicit alignment relies on accurate motion estimation, which can become unreliable under dense rain, while implicit aggregation avoids alignment but lacks explicit guidance on the directional and temporally coherent structure of rain. This leaves a gap between reliable temporal aggregation and explicit modeling of rain motion. To address these limitations, we propose FluidRain, a lightweight video derainer that uses divergence-free rain flow to guide Loop-in-Loop attention across scales and neighboring frames. Motivated by fluid mechanics, we model rain motion as a divergence-free image-space flow and use it to organize multi-scale and temporal aggregation. Specifically, FluidRain first estimates a rain-flow field for each frame and projects it onto the divergence-free subspace. The resulting flow steers window attention along rain streaks, enabling neighboring frames to be aggregated without explicit alignment. Since rain-flow structure is preserved across scales and nearby frames, Loop-in-Loop reuses the same attention operator across both dimensions, resulting in a three-frame model with only 0.80M parameters. Experiments on four benchmarks show that FluidRain remains competitive with substantially larger restoration models. We further examine how temporal evidence scales with different input views. To evaluate whether the model remains reliable when rain motion changes across frames, we introduce RainSyn-Gust, which injects controlled changes in rain-streak direction into existing benchmarks. We also develop a physics-based no-reference metric that evaluates real-rain removal without requiring clean targets.
Pu Wang, Yongcong Wang, Wenhao Li +6
Shandong University · Central South University · Nanjing University of Science and Technology +2
This paper presents an overview of the NTIRE 2026 Challenge on Short-form UGC Video Restoration in the Wild with Generative Models. This challenge utilizes a new short-form UGC (S-UGC) video restoration benchmark, termed KwaiVIR, which is contributed by USTC and Kuaishou Technology. It contains both synthetically distorted videos and real-world short-form UGC videos in the wild. For this edition, the released data include 200 synthetic training videos, 48 wild training videos, 11 validation videos, and 20 testing videos. The primary goal of this challenge is to establish a strong and practical benchmark for restoring short-form UGC videos under complex real-world degradations, especially in the emerging paradigm of generative-model-based S-UGC video restoration. This challenge has two tracks: (i) the primary track is a subjective track, where the evaluation is based on a user study; (ii) the second track is an objective track. These two tracks enable a comprehensive assessment of restoration quality. In total, 95 teams have registered for this competition. And 12 teams submitted valid final solutions and fact sheets for the testing phase. The submitted methods achieved strong performance on the KwaiVIR benchmark, demonstrating encouraging progress in short-form UGC video restoration in the wild.