We show that, simply initializing image understanding models using a pre-trained UNet (or transformer) of diffusion models, it is possible to achieve remarkable transferable performance on fundamental vision perception tasks using a moderate amount of target data (even synthetic data only), including monocular depth, surface normal, image segmentation, matting, human pose estimation, among virtually many others. Previous works have adapted diffusion models for various perception tasks, often reformulating these tasks as generation processes to align with the diffusion process. In sharp contrast, we demonstrate that fine-tuning these models with minimal adjustments can be a more effective alternative, offering the advantages of being embarrassingly simple and significantly faster. As the backbone network of Stable Diffusion models is trained on giant datasets comprising billions of images, we observe very robust generalization capabilities of the diffusion backbone. Experimental results showcase the remarkable transferability of the backbone of diffusion models across diverse tasks and real-world datasets.
3D scene reconstruction from 2D images has been a long-standing task. Instead of estimating per-frame depth maps and fusing them in 3D, recent research leverages the neural implicit surface as a unified representation for 3D reconstruction. Equipped with data-driven pre-trained geometric cues, these methods have demonstrated promising performance. However, inaccurate prior estimation, which is usually inevitable, can lead to suboptimal reconstruction quality, particularly in some geometrically complex regions. In this paper, we propose a two-stage training process, decouple view-dependent and view-independent colors, and leverage two novel consistency constraints to enhance detail reconstruction performance without requiring extra priors. Additionally, we introduce an essential mask scheme to adaptively influence the selection of supervision constraints, thereby improving performance in a self-supervised paradigm. Experiments on synthetic and real-world datasets show the capability of reducing the interference from prior estimation errors and achieving high-quality scene reconstruction with rich geometric details.
3D scene reconstruction is a long-standing vision task. Existing approaches can be categorized into geometry-based and learning-based methods. The former leverages multi-view geometry but can face catastrophic failures due to the reliance on accurate pixel correspondence across views. The latter was proffered to mitigate these issues by learning 2D or 3D representation directly. However, without a large-scale video or 3D training data, it can hardly generalize to diverse real-world scenarios due to the presence of tens of millions or even billions of optimization parameters in the deep network. Recently, robust monocular depth estimation models trained with large-scale datasets have been proven to possess weak 3D geometry prior, but they are insufficient for reconstruction due to the unknown camera parameters, the affine-invariant property, and inter-frame inconsistency. Here, we propose a novel test-time optimization approach that can transfer the robustness of affine-invariant depth models such as LeReS to challenging diverse scenes while ensuring inter-frame consistency, with only dozens of parameters to optimize per video frame. Specifically, our approach involves freezing the pre-trained affine-invariant depth model's depth predictions, rectifying them by optimizing the unknown scale-shift values with a geometric consistency alignment module, and employing the resulting scale-consistent depth maps to robustly obtain camera poses and achieve dense scene reconstruction, even in low-texture regions. Experiments show that our method achieves state-of-the-art cross-dataset reconstruction on five zero-shot testing datasets.
This paper discusses the results for the second edition of the Monocular Depth Estimation Challenge (MDEC). This edition was open to methods using any form of supervision, including fully-supervised, self-supervised, multi-task or proxy depth. The challenge was based around the SYNS-Patches dataset, which features a wide diversity of environments with high-quality dense ground-truth. This includes complex natural environments, e.g. forests or fields, which are greatly underrepresented in current benchmarks. The challenge received eight unique submissions that outperformed the provided SotA baseline on any of the pointcloud- or image-based metrics. The top supervised submission improved relative F-Score by 27.62%, while the top self-supervised improved it by 16.61%. Supervised submissions generally leveraged large collections of datasets to improve data diversity. Self-supervised submissions instead updated the network architecture and pretrained backbones. These results represent a significant progress in the field, while highlighting avenues for future research, such as reducing interpolation artifacts at depth boundaries, improving self-supervised indoor performance and overall natural image accuracy.
Transformer-based detectors (DETRs) have attracted great attention due to their sparse training paradigm and the removal of post-processing operations, but the huge model can be computationally time-consuming and difficult to be deployed in real-world applications. To tackle this problem, knowledge distillation (KD) can be employed to compress the huge model by constructing a universal teacher-student learning framework. Different from the traditional CNN detectors, where the distillation targets can be naturally aligned through the feature map, DETR regards object detection as a set prediction problem, leading to an unclear relationship between teacher and student during distillation. In this paper, we propose DETRDistill, a novel knowledge distillation dedicated to DETR-families. We first explore a sparse matching paradigm with progressive stage-by-stage instance distillation. Considering the diverse attention mechanisms adopted in different DETRs, we propose attention-agnostic feature distillation module to overcome the ineffectiveness of conventional feature imitation. Finally, to fully leverage the intermediate products from the teacher, we introduce teacher-assisted assignment distillation, which uses the teacher's object queries and assignment results for a group with additional guidance. Extensive experiments demonstrate that our distillation method achieves significant improvement on various competitive DETR approaches, without introducing extra consumption in the inference phase. To the best of our knowledge, this is the first systematic study to explore a general distillation method for DETR-style detectors.
Multi-view depth estimation plays a critical role in reconstructing and understanding the 3D world. Recent learning-based methods have made significant progress in it. However, multi-view depth estimation is fundamentally a correspondence-based optimization problem, but previous learning-based methods mainly rely on predefined depth hypotheses to build correspondence as the cost volume and implicitly regularize it to fit depth prediction, deviating from the essence of iterative optimization based on stereo correspondence. Thus, they suffer unsatisfactory precision and generalization capability. In this paper, we are the first to explore more general image correlations to establish correspondences dynamically for depth estimation. We design a novel iterative multi-view depth estimation framework mimicking the optimization process, which consists of 1) a correlation volume construction module that models the pixel similarity between a reference image and source images as all-to-all correlations; 2) a flow-based depth initialization module that estimates the depth from the 2D optical flow; 3) a novel correlation-guided depth refinement module that reprojects points in different views to effectively fetch relevant correlations for further fusion and integrate the fused correlation for iterative depth update. Without predefined depth hypotheses, the fused correlations establish multi-view correspondence in an efficient way and guide the depth refinement heuristically. We conduct sufficient experiments on ScanNet, DeMoN, ETH3D, and 7Scenes to demonstrate the superiority of our method on multi-view depth estimation and its best generalization ability.
Existing monocular depth estimation shows excellent robustness in the wild, but the affine-invariant prediction requires aligning with the ground truth globally while being converted into the metric depth. In this work, we firstly propose a modified locally weighted linear regression strategy to leverage sparse ground truth and generate a flexible depth transformation to correct the coarse misalignment brought by global recovery strategy. Applying this strategy, we achieve significant improvement (more than 50% at most) over most recent state-of-the-art methods on five zero-shot datasets. Moreover, we train a robust depth estimation model with 6.3 million data and analyze the training process by decoupling the inaccuracy into coarse misalignment inaccuracy and detail missing inaccuracy. As a result, our model based on ResNet50 even outperforms the state-of-the-art DPT ViT-Large model with the help of our recovery strategy. In addition to accuracy, the consistency is also boosted for simple per-frame video depth estimation. Compared with monocular depth estimation, robust video depth estimation, and depth completion methods, our pipeline obtains state-of-the-art performance on video depth estimation without any post-processing. Experiments of 3D scene reconstruction from consistent video depth are conducted for intuitive comparison as well.