Ziteng Cui

Ziteng Cui (崔子藤)

I'm a Ph.D. student at The University of Tokyo, where I am supervised by Prof. Tatsuya Harada. Before that, I received my master's degree from Shanghai Jiao Tong University.

I mainly work on Vision Robustness & Computational Photography & 3D Computer Vision. My favorite things are DOTA2, hiking, JOJO (Jolyne Cujoh) and Pink Floyd.

I am graduating in April 2025 and am seeking opportunities in both academia and industry. Please feel free to connect with me via email.

Email / Google Scholar / Github / Twitter / LinkedIn

昔者庄周梦为胡蝶，栩栩然胡蝶也，自喻适志与，不知周也。俄然觉，则蘧蘧然周也。不知周之梦为胡蝶与？胡蝶之梦为周与？周与胡蝶，则必有分矣，此之谓物化。 -- 庄子《齐物论》

News

● 2025.6.01 I successfully defended my Ph.D. dissertation, Dr. Cui now 💐💐
● 2025.4.12 I have been selected to the Doctoral Consortium at CVPR 2025 ☀️☀️
● 2025.2.27 One first author paper Luminance-GS has been accepted by CVPR 2025 ☀️☀️
● 2024.7.01 One first author paper RAW-Adapter has been accepted by ECCV 2024 ☀️☀️
● 2023.12.09 One first author paper Aleth-NeRF has been accepted by AAAI 2024 ☀️☀️
● ... ...

Selected Publications

I'm now interested in the physics modeling of low-level vision and 3D computer vision, I especially interested in the neural radiance field. "*" means authors contribute equally. Full publication list please refer to here.

	Luminance-GS: Adapting 3D Gaussian Splatting to Challenging Lighting Conditions with View-Adaptive Curve Adjustment Ziteng Cui, Xuangeng Chu, Tatsuya Harada. CVPR, 2025 [github] website / arxiv / bibtex / poster A simple multi-view curve adjustment method for novel view synthesis under challenging lighting conditions, including low-light, overexposure, and varying exposure.
	ARTalk: Speech-Driven 3D Head Animation via Autoregressive Model Xuangeng Chu, Nabarun Goswami, Ziteng Cui, Hanqin Wang, Tatsuya Harada. Arxiv, 2025 [github] website / arxiv / bibtex ARTalk generates realistic 3D head motions (lip sync, blinking, expressions, head poses) from audio in real-time.
	Discovering an Image-Adaptive Coordinate System for Photography Processing Ziteng Cui, Lin Gu, Tatsuya Harada. BMVC, 2024 arxiv / bibtex Instead of just using image-adaptive curves or 3D LUTs, why not design an image-adaptive coordinate system tailored to different photography processing tasks?
	RAW-Adapter: Adapting Pre-trained Visual Model to Camera RAW Images Ziteng Cui, Tatsuya Harada. ECCV, 2024 [github] website / arxiv / bibtex / poster We analyze the relationship between camera RAW data and sRGB images pre-trained models, and propose RAW-Adapter for effective RAW data based vision tasks.
	Aleth-NeRF: Illumination Adaptive NeRF with Concealing Field Assumption Ziteng Cui, Lin Gu, Xiao Sun, Xianzheng Ma, Yu Qiao, Tatsuya Harada. AAAI, 2024 [github] website / arxiv / arxiv (old version) / bibtex / poster Blight NeRF's volume rendering function with concealing fields, to handle novel view synthesis under low-light conditions and overexposure conditions.
	Monodetr: Depth-guided transformer for monocular 3d object detection Renrui Zhang, Han Qiu, Tai Wang, Ziyu Guo, Ziteng Cui, Yu Qiao, Hongsheng Li, Peng Gao ICCV, 2023 [github] arxiv / bibtex / code / poster MonoDETR introduces a depth-guided transformer for monocular 3D object detection, enhancing Mono3D with non-local depth cues and achieving SOTA results.
	You Only Need 90K Parameters to Adapt Light: A Light Weight Transformer for Image Enhancement and Exposure Correction Ziteng Cui, Kunchang Li, Lin Gu, Shenghan Su, Peng Gao, Zhengkai Jiang, Yu Qiao, Tatsuya Harada. BMVC, 2022 [github] arxiv / bibtex / demo / poster A super light-weight (only 90k+ parameters) transformer-based network Illumination Adaptive Transformer, for real time image enhancement and exposure correction.
	Exploring Resolution and Degradation Clues as Self-supervised Signal for Low Quality Object Detection Ziteng Cui, Yingying Zhu, Lin Gu, Guo-Jun Qi, Xiaoxiao Li, Renrui Zhang, Zenghui Zhang, Tatsuya Harada. ECCV, 2022 [github] arxiv / bibtex / poster Combine detection with self-supervised super-resolution, for robust detection under various degradation conditions (noise, blurry, low-resolution).
	Multitask AET with Orthogonal Tangent Regularity for Dark Object Detection Ziteng Cui, Guo-Jun Qi, Lin Gu, Shaodi You, Zenghui Zhang, Tatsuya Harada. ICCV, 2021 [github] arxiv / bibtex / poster Using Camera-ISP pipeline for low-light image synthetic, then using self-supervised learning to improving the performance of low-light condition object detection.