Shizhan Zhu

Rodolfo Corona, Shizhan Zhu, Dan Klein et al.

Natural language applied to natural 2D images describes a fundamentally 3D world. We present the Voxel-informed Language Grounder (VLG), a language grounding model that leverages 3D geometric information in the form of voxel maps derived from the visual input using a volumetric reconstruction model. We show that VLG significantly improves grounding accuracy on SNARE, an object reference game task. At the time of writing, VLG holds the top place on the SNARE leaderboard, achieving SOTA results with a 2.0% absolute improvement.

Shizhan Zhu, Shunsuke Saito, Aljaz Bozic et al.

Image-based lighting (IBL) is a widely used technique that renders objects using a high dynamic range image or environment map. However, aggregating the irradiance at the object's surface is computationally expensive, in particular for non-opaque, translucent materials that require volumetric rendering techniques. In this paper we present a fast neural 3D reconstruction and relighting model that extends volumetric implicit models such as neural radiance fields to be relightable using IBL. It is general enough to handle materials that exhibit complex light transport effects, such as translucency and glossy reflections from detailed surface geometry, producing realistic and compelling results. Rendering can be within a second at 800$\times$800 resolution (0.72s on an NVIDIA 3090 GPU and 0.30s on an A100 GPU) without engineering optimization. Our code and dataset are available at https://zhusz.github.io/TRHM-Webpage/.

Daniel C. Moura, Shizhan Zhu, Orly Zvitia

This paper presents the Nexar Dashcam Collision Prediction Dataset and Challenge, designed to support research in traffic event analysis, collision prediction, and autonomous vehicle safety. The dataset consists of 1,500 annotated video clips, each approximately 40 seconds long, capturing a diverse range of real-world traffic scenarios. Videos are labeled with event type (collision/near-collision vs. normal driving), environmental conditions (lighting conditions and weather), and scene type (urban, rural, highway, etc.). For collision and near-collision cases, additional temporal labels are provided, including the precise moment of the event and the alert time, marking when the collision first becomes predictable. To advance research on accident prediction, we introduce the Nexar Dashcam Collision Prediction Challenge, a public competition on top of this dataset. Participants are tasked with developing machine learning models that predict the likelihood of an imminent collision, given an input video. Model performance is evaluated using the average precision (AP) computed across multiple intervals before the accident (i.e. 500 ms, 1000 ms, and 1500 ms prior to the event), emphasizing the importance of early and reliable predictions. The dataset is released under an open license with restrictions on unethical use, ensuring responsible research and innovation.

Roni Goldshmidt, Hamish Scott, Lorenzo Niccolini et al.

Existing collision prediction methods often fail to distinguish between ego-vehicle threats and random accidents not involving the ego vehicle, leading to excessive false alerts in real-world deployment. We present BADAS, a family of collision prediction models trained on Nexar's real-world dashcam collision dataset -- the first benchmark designed explicitly for ego-centric evaluation. We re-annotate major benchmarks to identify ego involvement, add consensus alert-time labels, and synthesize negatives where needed, enabling fair AP/AUC and temporal evaluation. BADAS uses a V-JEPA2 backbone trained end-to-end and comes in two variants: BADAS-Open (trained on our 1.5k public videos) and BADAS1.0 (trained on 40k proprietary videos). Across DAD, DADA-2000, DoTA, and Nexar, BADAS achieves state-of-the-art AP/AUC and outperforms a forward-collision ADAS baseline while producing more realistic time-to-accident estimates. We release our BADAS-Open model weights and code, along with re-annotations of all evaluation datasets to promote ego-centric collision prediction research.

Sayna Ebrahimi, William Gan, Dian Chen et al.

Active learning aims to develop label-efficient algorithms by querying the most representative samples to be labeled by a human annotator. Current active learning techniques either rely on model uncertainty to select the most uncertain samples or use clustering or reconstruction to choose the most diverse set of unlabeled examples. While uncertainty-based strategies are susceptible to outliers, solely relying on sample diversity does not capture the information available on the main task. In this work, we develop a semi-supervised minimax entropy-based active learning algorithm that leverages both uncertainty and diversity in an adversarial manner. Our model consists of an entropy minimizing feature encoding network followed by an entropy maximizing classification layer. This minimax formulation reduces the distribution gap between the labeled/unlabeled data, while a discriminator is simultaneously trained to distinguish the labeled/unlabeled data. The highest entropy samples from the classifier that the discriminator predicts as unlabeled are selected for labeling. We evaluate our method on various image classification and semantic segmentation benchmark datasets and show superior performance over the state-of-the-art methods.

Shizhan Zhu, Sanja Fidler, Raquel Urtasun et al.

We present a novel and effective approach for generating new clothing on a wearer through generative adversarial learning. Given an input image of a person and a sentence describing a different outfit, our model "redresses" the person as desired, while at the same time keeping the wearer and her/his pose unchanged. Generating new outfits with precise regions conforming to a language description while retaining wearer's body structure is a new challenging task. Existing generative adversarial networks are not ideal in ensuring global coherence of structure given both the input photograph and language description as conditions. We address this challenge by decomposing the complex generative process into two conditional stages. In the first stage, we generate a plausible semantic segmentation map that obeys the wearer's pose as a latent spatial arrangement. An effective spatial constraint is formulated to guide the generation of this semantic segmentation map. In the second stage, a generative model with a newly proposed compositional mapping layer is used to render the final image with precise regions and textures conditioned on this map. We extended the DeepFashion dataset [8] by collecting sentence descriptions for 79K images. We demonstrate the effectiveness of our approach through both quantitative and qualitative evaluations. A user study is also conducted. The codes and the data are available at http://mmlab.ie.cuhk. edu.hk/projects/FashionGAN/.

Shizhan Zhu, Sifei Liu, Chen Change Loy et al.

We present a novel framework for hallucinating faces of unconstrained poses and with very low resolution (face size as small as 5pxIOD). In contrast to existing studies that mostly ignore or assume pre-aligned face spatial configuration (e.g. facial landmarks localization or dense correspondence field), we alternatingly optimize two complementary tasks, namely face hallucination and dense correspondence field estimation, in a unified framework. In addition, we propose a new gated deep bi-network that contains two functionality-specialized branches to recover different levels of texture details. Extensive experiments demonstrate that such formulation allows exceptional hallucination quality on in-the-wild low-res faces with significant pose and illumination variations.

Shizhan Zhu, Cheng Li, Chen Change Loy et al.

Learning to simultaneously handle face alignment of arbitrary views, e.g. frontal and profile views, appears to be more challenging than we thought. The difficulties lay in i) accommodating the complex appearance-shape relations exhibited in different views, and ii) encompassing the varying landmark point sets due to self-occlusion and different landmark protocols. Most existing studies approach this problem via training multiple viewpoint-specific models, and conduct head pose estimation for model selection. This solution is intuitive but the performance is highly susceptible to inaccurate head pose estimation. In this study, we address this shortcoming through learning an Ensemble of Model Recommendation Trees (EMRT), which is capable of selecting optimal model configuration without prior head pose estimation. The unified framework seamlessly handles different viewpoints and landmark protocols, and it is trained by optimising directly on landmark locations, thus yielding superior results on arbitrary-view face alignment. This is the first study that performs face alignment on the full AFLWdataset with faces of different views including profile view. State-of-the-art performances are also reported on MultiPIE and AFW datasets containing both frontaland profile-view faces.

Shizhan Zhu, Cheng Li, Chen Change Loy et al.

Dataset bias is a well known problem in object recognition domain. This issue, nonetheless, is rarely explored in face alignment research. In this study, we show that dataset plays an integral part of face alignment performance. Specifically, owing to face alignment dataset bias, training on one database and testing on another or unseen domain would lead to poor performance. Creating an unbiased dataset through combining various existing databases, however, is non-trivial as one has to exhaustively re-label the landmarks for standardisation. In this work, we propose a simple and yet effective method to bridge the disparate annotation spaces between databases, making datasets fusion possible. We show extensive results on combining various popular databases (LFW, AFLW, LFPW, HELEN) for improved cross-dataset and unseen data alignment.