no code implementations • 6 Jan 2024 • ChungYi Lin, Shen-Lung Tung, Hung-Ting Su, Winston H. Hsu
To address the limitations of traffic prediction from location-bound detectors, we present Geographical Cellular Traffic (GCT) flow, a novel data source that leverages the extensive coverage of cellular traffic to capture mobility patterns.
1 code implementation • 5 Oct 2023 • Tsung-Lin Tsou, Tsung-Han Wu, Winston H. Hsu
In the field of domain adaptation (DA) on 3D object detection, most of the work is dedicated to unsupervised domain adaptation (UDA).
1 code implementation • 7 Aug 2023 • Chien Cheng Chyou, Hung-Ting Su, Winston H. Hsu
Adversarial robustness poses a critical challenge in the deployment of deep learning models for real-world applications.
no code implementations • 7 Apr 2023 • Hung-Ting Su, Yulei Niu, Xudong Lin, Winston H. Hsu, Shih-Fu Chang
Causal Video Question Answering (CVidQA) queries not only association or temporal relations but also causal relations in a video.
no code implementations • 29 Mar 2023 • Yi-Syuan Liou, Tsung-Han Wu, Jia-Fong Yeh, Wen-Chin Chen, Winston H. Hsu
MuRAL identifies informative regions of various scales to reduce annotation costs for well-learned objects and improve training performance.
1 code implementation • 16 Dec 2022 • Ru-Fen Jheng, Tsung-Han Wu, Jia-Fong Yeh, Winston H. Hsu
Thus, we present a novel task named free-form 3D scene inpainting.
1 code implementation • 8 Oct 2022 • Hsin-Ying Lee, Hung-Ting Su, Bing-Chen Tsai, Tsung-Han Wu, Jia-Fong Yeh, Winston H. Hsu
While recent large-scale video-language pre-training made great progress in video question answering, the design of spatial modeling of video-language models is less fine-grained than that of image-language models; existing practices of temporal modeling also suffer from weak and noisy alignment between modalities.
1 code implementation • 5 Oct 2022 • Cheng-Wei Lin, Tung-I Chen, Hsin-Ying Lee, Wen-Chin Chen, Winston H. Hsu
As global feature alignment requires the features to preserve the poses of input point clouds and local feature matching expects the features to be invariant to these poses, we propose an SE(3)-equivariant feature extractor to simultaneously generate two types of features.
1 code implementation • 27 Sep 2022 • Ching-Yu Tseng, Yi-Rong Chen, Hsin-Ying Lee, Tsung-Han Wu, Wen-Chin Chen, Winston H. Hsu
To achieve accurate 3D object detection at a low cost for autonomous driving, many multi-camera methods have been proposed and solved the occlusion problem of monocular approaches.
1 code implementation • 27 Sep 2022 • Chi-Ming Chung, Yang-Che Tseng, Ya-Ching Hsu, Xiang-Qian Shi, Yun-Hung Hua, Jia-Fong Yeh, Wen-Chin Chen, Yi-Ting Chen, Winston H. Hsu
A spatial AI that can perform complex tasks through visual signals and cooperate with humans is highly anticipated.
no code implementations • 22 Sep 2022 • Tsung-Han Wu, Hung-Ting Su, Shang-Tse Chen, Winston H. Hsu
Fairness and robustness play vital roles in trustworthy machine learning.
2 code implementations • 7 May 2022 • Igor Morawski, Yu-An Chen, Yu-Sheng Lin, Shusil Dangi, Kai He, Winston H. Hsu
We propose to improve generalization to unseen camera sensors by implementing a minimal neural ISP pipeline for machine cognition, named GenISP, that explicitly incorporates Color Space Transformation to a device-independent color space.
1 code implementation • CVPR 2022 • Kuan-Chih Huang, Tsung-Han Wu, Hung-Ting Su, Winston H. Hsu
Moreover, different from conventional pixel-wise positional encodings, we introduce a novel depth positional encoding (DPE) to inject depth positional hints into transformers.
1 code implementation • 14 Feb 2022 • Tsung-Han Wu, Yi-Syuan Liou, Shao-Ji Yuan, Hsin-Ying Lee, Tung-I Chen, Kuan-Chih Huang, Winston H. Hsu
In the field of domain adaptation, a trade-off exists between the model performance and the number of target domain annotations.
no code implementations • 4 Dec 2021 • Jia-Fong Yeh, Chi-Ming Chung, Hung-Ting Su, Yi-Ting Chen, Winston H. Hsu
(3) Learning from a different expert.
no code implementations • 2 Dec 2021 • Ching-Yu Tseng, Po-Shao Lin, Yu-Jia Liou, Kuan-Chih Huang, Winston H. Hsu
Shifts Challenge: Robustness and Uncertainty under Real-World Distributional Shift is a competition held by NeurIPS 2021.
no code implementations • 29 Nov 2021 • Guan-Rong Lu, Yueh-Cheng Liu, Tung-I Chen, Hung-Ting Su, Tsung-Han Wu, Winston H. Hsu
We design a new Masked Gradient Update (MGU) module to generate auxiliary data along the boundary of in-distribution data points.
no code implementations • 22 Oct 2021 • Kuan-Chih Huang, Yu-Kai Huang, Winston H. Hsu
Vehicle velocity and inter-vehicle distance estimation are essential for ADAS (Advanced driver-assistance systems) and autonomous vehicles.
1 code implementation • 20 Oct 2021 • Igor Morawski, Yu-An Chen, Yu-Sheng Lin, Winston H. Hsu
In our work, we take a closer look at object detection in low light.
1 code implementation • 18 Aug 2021 • Chung-Yi Lin, Hung-Ting Su, Shen-Lung Tung, Winston H. Hsu
Furthermore, we propose a new model for multivariate spatial-temporal prediction, mainly consisting of two extending graph attention networks (GAT).
no code implementations • 10 Aug 2021 • Hung-Ting Su, Po-Wei Shen, Bing-Chen Tsai, Wen-Feng Cheng, Ke-Jyun Wang, Winston H. Hsu
By coping with the trope understanding task and enabling the deep cognition skills of machines, data mining applications and algorithms could be taken to the next level.
1 code implementation • ICCV 2021 • Tsung-Han Wu, Yueh-Cheng Liu, Yu-Kai Huang, Hsin-Ying Lee, Hung-Ting Su, Ping-Chia Huang, Winston H. Hsu
Despite the success of deep learning on supervised point cloud semantic segmentation, obtaining large-scale point-by-point manual annotations is still a significant challenge.
no code implementations • CVPR 2021 • Yu-Kai Huang, Yueh-Cheng Liu, Tsung-Han Wu, Hung-Ting Su, Yu-Cheng Chang, Tsung-Lin Tsou, Yu-An Wang, Winston H. Hsu
Dense depth estimation plays a key role in multiple applications such as robotics, 3D reconstruction, and augmented reality.
1 code implementation • WACV 2021 • Shuo-Diao Yang, Hung-Ting Su, Winston H. Hsu, Wen-Chin Chen
Instead of counting a pre-defined class, our model is able to count instances based on input reference images and reduces the huge cost of data collection, training and parameter tuning for each new object class.
no code implementations • 10 Apr 2021 • Yueh-Cheng Liu, Yu-Kai Huang, Hung-Yueh Chiang, Hung-Ting Su, Zhe-Yu Liu, Chin-Tang Chen, Ching-Yu Tseng, Winston H. Hsu
Most 3D neural networks are trained from scratch owing to the lack of large-scale labeled 3D datasets.
1 code implementation • NAACL 2021 • Ke-Jyun Wang, Yun-Hsuan Liu, Hung-Ting Su, Jen-Wei Wang, Yu-Siang Wang, Winston H. Hsu, Wen-Chin Chen
To effectively apply robots in working environments and assist humans, it is essential to develop and evaluate how visual grounding (VG) can affect machine performance on occluded objects.
no code implementations • 3 Mar 2021 • Yu-Kai Huang, Yueh-Cheng Liu, Tsung-Han Wu, Hung-Ting Su, Yu-Cheng Chang, Tsung-Lin Tsou, Yu-An Wang, Winston H. Hsu
Dense depth estimation plays a key role in multiple applications such as robotics, 3D reconstruction, and augmented reality.
1 code implementation • 24 Feb 2021 • Tung-I Chen, Yueh-Cheng Liu, Hung-Ting Su, Yu-Cheng Chang, Yu-Hsiang Lin, Jia-Fong Yeh, Wen-Chin Chen, Winston H. Hsu
While recent progress has significantly boosted few-shot classification (FSC) performance, few-shot object detection (FSOD) remains challenging for modern learning systems.
Ranked #9 on Few-Shot Object Detection on MS-COCO (10-shot)
1 code implementation • 19 Jan 2021 • Chen-Hsi Chang, Hung-Ting Su, Jui-heng Hsu, Yu-Siang Wang, Yu-Cheng Chang, Zhe Yu Liu, Ya-Liang Chang, Wen-Feng Cheng, Ke-Jyun Wang, Winston H. Hsu
Experimental result demonstrates that modern models including BERT contextual embedding, movie tag prediction systems, and relational networks, perform at most 37% of human performance (23. 97/64. 87) in terms of F1 score.
1 code implementation • 5 Jan 2021 • Hung-Ting Su, Chen-Hsi Chang, Po-Wei Shen, Yu-Siang Wang, Ya-Liang Chang, Yu-Cheng Chang, Pu-Jen Cheng, Winston H. Hsu
Furthermore, using our generated QA pairs only on the Video QA task, we can surpass some supervised baselines.
1 code implementation • 8 Dec 2020 • Chih-Hung Liang, Yu-An Chen, Yueh-Cheng Liu, Winston H. Hsu
Therefore, we built a new dataset containing both RAW images and processed sRGB images and design a new model to utilize the unique characteristics of RAW images.
no code implementations • 21 Oct 2020 • Kuang-Yu Jeng, Yueh-Cheng Liu, Zhe Yu Liu, Jen-Wei Wang, Ya-Liang Chang, Hung-Ting Su, Winston H. Hsu
We proposed an end-to-end grasp detection network, Grasp Detection Network (GDN), cooperated with a novel coarse-to-fine (C2F) grasp representation design to detect diverse and accurate 6-DoF grasps based on point clouds.
no code implementations • 21 May 2020 • Jhih-Yuan Lin, Yu-Cheng Chang, Winston H. Hsu
Cardiac Magnetic Resonance Imaging (CMR) is widely used since it can illustrate the structure and function of heart in a non-invasive and painless way.
no code implementations • 19 May 2020 • Jia-Fong Yeh, Hsin-Ying Lee, Bing-Chen Tsai, Yi-Rong Chen, Ping-Chia Huang, Winston H. Hsu
In recent years, few-shot learning problems have received a lot of attention.
no code implementations • 24 Apr 2020 • Yu-Kai Huang, Yueh-Cheng Liu, Tsung-Han Wu, Hung-Ting Su, Winston H. Hsu
The performance of image based stereo estimation suffers from lighting variations, repetitive patterns and homogeneous appearance.
1 code implementation • 11 Mar 2020 • Yu-Sheng Lin, Zhe-Yu Liu, Yu-An Chen, Yu-Siang Wang, Ya-Liang Chang, Winston H. Hsu
We study the XAI (explainable AI) on the face recognition task, particularly the face verification here.
Explainable Artificial Intelligence (XAI) Face Recognition +1
3 code implementations • 22 Aug 2019 • Yu-Kai Huang, Tsung-Han Wu, Yueh-Cheng Liu, Winston H. Hsu
We utilize self-attention mechanism, previously used in image inpainting fields, to extract more useful information in each layer of convolution so that the complete depth map is enhanced.
Ranked #2 on Depth Completion on Matterport3D
1 code implementation • 1 Aug 2019 • Hung-Yueh Chiang, Yen-Liang Lin, Yueh-Cheng Liu, Winston H. Hsu
We present a new unified point-based framework for 3D point cloud segmentation that effectively optimizes pixel-level features, geometrical structures and global context priors of an entire scene.
no code implementations • 30 Jul 2019 • Sebastian Agethen, Winston H. Hsu
Herein, we propose a new enhancement to convolutional LSTM networks that supports accommodation of multiple convolutional kernels and layers.
no code implementations • 5 Jul 2019 • Yu-Siang Wang, Hung-Ting Su, Chen-Hsi Chang, Zhe-Yu Liu, Winston H. Hsu
We introduce a novel task, Video Question Generation (Video QG).
no code implementations • ECCV 2018 • Kaipeng Zhang, Zhanpeng Zhang, Chia-Wen Cheng, Winston H. Hsu, Yu Qiao, Wei Liu, Tong Zhang
Face hallucination is a generative task to super-resolve the facial image with low resolution while human perception of face heavily relies on identity information.
no code implementations • ICCV 2017 • Meng-Ru Hsieh, Yen-Liang Lin, Winston H. Hsu
Existing counting methods often adopt regression-based approaches and cannot precisely localize the target objects, which hinders the further analysis (e. g., high-level understanding and fine-grained classification).
Ranked #8 on Object Counting on CARPK
no code implementations • 18 Aug 2016 • Wei-Tse Sun, Ting-Hsuan Chao, Yin-Hsi Kuo, Winston H. Hsu
We conduct experiments on the collected FACD for filter recommendation, and the results show that our proposed category-aware aesthetic learning outperforms aesthetic classification methods (e. g., 12% relative improvement).
no code implementations • 29 Jun 2016 • Yin-Hsi Kuo, Winston H. Hsu
Based on the hashed binary codes, we propose a de-hashing process that reconstructs BoW by leveraging the computing power of remote servers.
no code implementations • 19 Nov 2015 • Sebastian Agethen, Winston H. Hsu
Our architecture achieves this with the help of expert networks: A network is trained on a disjoint subset of a given dataset and then run in parallel to other experts during deployment.
no code implementations • 19 Aug 2015 • Kuan-Ting Chen, Kezhen Chen, Peizhong Cong, Winston H. Hsu, Jiebo Luo
To answer this question, we design a novel system that consists of three major components: (1) constructing a large dataset from the New York Fashion Shows and New York street chic in order to understand the likely clothing fashion trends in New York, (2) utilizing a learning-based approach to discover fashion attributes as the representative characteristics of fashion trends, and (3) comparing the analysis results from the New York Fashion Shows and street-chic images to verify whether the fashion shows have actual influence on the people in New York City.
no code implementations • CVPR 2015 • Ting-Hsuan Chao, Yen-Liang Lin, Yin-Hsi Kuo, Winston H. Hsu
Our method can reconstruct filters by minimizing score map error, while sparse coding reconstructs filters by minimizing appearance error.
no code implementations • 15 Sep 2014 • Yu-Chuan Su, Tzu-Hsuan Chiu, Chun-Yen Yeh, Hsin-Fu Huang, Winston H. Hsu
The same lack-of-training-sample problem limits the usage of deep models on a wide range of computer vision problems where obtaining training data are difficult.