Wenzhao Zheng

I am a fifth year Ph.D student in the Department of Automation at Tsinghua University, advised by Prof. Jie Zhou and Prof. Jiwen Lu . In 2018, I received my BS degree from the Department of Physics, Tsinghua University.

I am interested in computer vision and deep learning. My current research focuses on:

Email  /  CV  /  Google Scholar  /  GitHub

Publications

* indicates equal contribution

We propose an introspective deep metric learning (IDML) framework for uncertainty-aware comparisons of images.

We propose a Structure-Preserving Encoding (SPE) module to reformulate depth completion as a process of 3D structure generation.

We unify fully supervised and self-supervised contrastive learning and exploit both supervisions from labeled and unlabeled data for training.

We identify a token fluctuation phenomenon that has suppressed the potential of data mixing strategies for vision transformers. To adress this, we propose a token-label alignment (TL-Align) method to trace the correspondence between transformed tokens and the original tokens to maintain a label for each token.

We design a pipeline to generate dense occupancy ground truths without expensive occupancy annotations, which enalbes the training of more dense 3D occupancy prediction models.

Given only surround-camera motorcycle RGB images barrier as inputs, our model (trained using trailer only sparse traffic cone LiDAR point supervision) can predict the semantic occupancy for all volumes in the 3D space.

We factorize the backbone network to different sub-blocks and learns an adaptive route for each sample to achieve the diversity of features.

This paper proposes Bort, an optimizer for improving model explainability with boundedness and orthogonality constraints on model parameters, derived from the sufficient conditions of model comprehensibility and transparency.

This paper formulates deep metric learning under a unified framework and propose a dynamic constraint generator to produce adaptive composite constraints to train the metric towards good generalization.

We propose a probabilistic deep metric learning framework to model the categorical uncertainty of the spectral distribution of an observed pixel for Hyperspectral image classification.

This paper propose a hierarchical concept refiner to construct multiple levels of concept embeddings of an image and them pull closer the distance of the corresponding concepts to facilitate the cross-level semantic structure of the image representations.

We propose a simple baseline for multi-camera object detection to adapt existing monocular 3D object detection methods with a two-stage propose-and-fuse framework.

We propose a unified framework for 3D perception and prediction based on multi-camera systems. The multi-task BEVerse outperforms existing single-task methods on 3D object detection, semantic map construction, and motion prediction.

We propose a SurroundDepth method to incorporate the information from multiple surrounding views to predict depth maps across cameras.

We propose a general method to learn appropriate embeddings for dimension estimation in monocular 3D object detection.

This paper proposes an attributable visual similarity learning (AVSL) framework, which employs a generalized similarity learning paradigm to represent the similarity between two images with a graph for a more accurate and explainable similarity measure between images.

We construct a graph to represent each image and perform relational inference to infer the visual similarity.

We adaptively learn a set of composites of embeddings to receive supervision signals from different tasks to improve the generalization of the learned embeddings without sacrificing the discriminativeness.

We are the first to apply deep metric learning to prediction tasks with structured labels.

We extend the previous conference-verision HDML to generate multiple synthetics for each sample.

We learn a sample assessment strategy for deep metric learning to maximize the generalization of the trained metric.

We propose a deep adversarial multi-metric learning (DAMML) method by learning multiple local transformations for more complete description.

We perform linear interpolation on embeddings to adaptively manipulate their hardness levels and generate corresponding label-preserving synthetics for recycled training.

We generate potential hard negatives adversarial to the learned metric as complements.

Website Template