Search Results for author: Mengmi Zhang

Found 26 papers, 16 papers with code

Adaptive Visual Scene Understanding: Incremental Scene Graph Generation

no code implementations2 Oct 2023 Naitik Khandelwal, Xiao Liu, Mengmi Zhang

To address the lack of continual learning methodologies in SGG, we introduce the comprehensive Continual ScenE Graph Generation (CSEGG) dataset along with 3 learning scenarios and 8 evaluation metrics.

Integrating Curricula with Replays: Its Effects on Continual Learning

1 code implementation8 Jul 2023 Ren Jie Tee, Mengmi Zhang

Our study takes initial steps in examining the impact of integrating curricula with replay methods on continual learning in three specific aspects: the interleaved frequency of replayed exemplars with training data, the sequence in which exemplars are replayed, and the strategy for selecting exemplars into the replay buffer.

Continual Learning Transfer Learning

Training-free Object Counting with Prompts

1 code implementation30 Jun 2023 Zenglin Shi, Ying Sun, Mengmi Zhang

However, the vanilla mask generation method of SAM lacks class-specific information in the masks, resulting in inferior counting accuracy.

Object Counting Zero Shot Segmentation

Object-centric Learning with Cyclic Walks between Parts and Whole

no code implementations16 Feb 2023 Ziyu Wang, Mike Zheng Shou, Mengmi Zhang

To capture compositional entities of the scene, we proposed cyclic walks between perceptual features extracted from CNN or transformers and object entities.

Efficient Zero-shot Visual Search via Target and Context-aware Transformer

no code implementations24 Nov 2022 Zhiwei Ding, Xuezhe Ren, Erwan David, Melissa Vo, Gabriel Kreiman, Mengmi Zhang

Target modulation is computed as patch-wise local relevance between the target and search images, whereas contextual modulation is applied in a global fashion.

Reason from Context with Self-supervised Learning

no code implementations23 Nov 2022 Xiao Liu, Ankur Sikarwar, Gabriel Kreiman, Zenglin Shi, Mengmi Zhang

To better accommodate the object-centric nature of current downstream tasks such as object recognition and detection, various methods have been proposed to suppress contextual biases or disentangle objects from contexts.

Object Recognition Self-Supervised Learning +1

On the Robustness, Generalization, and Forgetting of Shape-Texture Debiased Continual Learning

no code implementations21 Nov 2022 Zenglin Shi, Ying Sun, Joo Hwee Lim, Mengmi Zhang

Tremendous progress has been made in continual learning to maintain good performance on old tasks when learning new tasks by tackling the catastrophic forgetting problem of neural networks.

Continual Learning

What makes domain generalization hard?

no code implementations15 Jun 2022 Spandan Madan, Li You, Mengmi Zhang, Hanspeter Pfister, Gabriel Kreiman

Here we present SemanticDG (Semantic Domain Generalization): a benchmark with 15 photo-realistic domains with the same geometry, scene layout and camera parameters as the popular 3D ScanNet dataset, but with controlled domain shifts in lighting, materials, and viewpoints.

Domain Generalization

Label-Efficient Online Continual Object Detection in Streaming Video

1 code implementation ICCV 2023 Jay Zhangjie Wu, David Junhao Zhang, Wynne Hsu, Mengmi Zhang, Mike Zheng Shou

Remarkably, with only 25% annotated video frames, our method still outperforms the base CL learners, which are trained with 100% annotations on all video frames.

Continual Learning Hippocampus +2

Visual Search Asymmetry: Deep Nets and Humans Share Similar Inherent Biases

1 code implementation NeurIPS 2021 Shashi Kant Gupta, Mengmi Zhang, Chia-Chien Wu, Jeremy M. Wolfe, Gabriel Kreiman

To elucidate the mechanisms responsible for asymmetry in visual search, we propose a computational model that takes a target and a search image as inputs and produces a sequence of eye movements until the target is found.

Tuned Compositional Feature Replays for Efficient Stream Learning

1 code implementation6 Apr 2021 Morgan B. Talbot, Rushikesh Zawar, Rohil Badkundri, Mengmi Zhang, Gabriel Kreiman

CRUMB's memory blocks are tuned to enhance replay: a single feature map stored, reconstructed, and replayed by CRUMB mitigates forgetting during video stream learning more effectively than an entire image, even though it occupies only 3. 6% as much memory.

Continual Learning Image Classification +2

When Pigs Fly: Contextual Reasoning in Synthetic and Natural Scenes

1 code implementation ICCV 2021 Philipp Bomatter, Mengmi Zhang, Dimitar Karev, Spandan Madan, Claire Tseng, Gabriel Kreiman

Our model captures useful information for contextual reasoning, enabling human-level performance and better robustness in out-of-context conditions compared to baseline models across OCD and other out-of-context datasets.

What am I Searching for: Zero-shot Target Identity Inference in Visual Search

1 code implementation25 May 2020 Mengmi Zhang, Gabriel Kreiman

Using those error fixations, we developed a model (InferNet) to infer what the target was.

Putting visual object recognition in context

1 code implementation CVPR 2020 Mengmi Zhang, Claire Tseng, Gabriel Kreiman

To model the role of contextual information in visual recognition, we systematically investigated ten critical properties of where, when, and how context modulates recognition, including the amount of context, context and object resolution, geometrical structure of context, context congruence, and temporal dynamics of contextual modulation.

Object Recognition

Prototype Recalls for Continual Learning

no code implementations25 Sep 2019 Mengmi Zhang, Tao Wang, Joo Hwee Lim, Jiashi Feng

Without tampering with the performance on initial tasks, our method learns novel concepts given a few training examples of each class in new tasks.

Continual Learning Metric Learning +1

Variational Prototype Replays for Continual Learning

1 code implementation23 May 2019 Mengmi Zhang, Tao Wang, Joo Hwee Lim, Gabriel Kreiman, Jiashi Feng

In each classification task, our method learns a set of variational prototypes with their means and variances, where embedding of the samples from the same class can be represented in a prototypical distribution and class-representative prototypes are separated apart.

Continual Learning General Classification +2

Lift-the-flap: what, where and when for context reasoning

no code implementations1 Feb 2019 Mengmi Zhang, Claire Tseng, Karla Montejo, Joseph Kwon, Gabriel Kreiman

Context reasoning is critical in a wide variety of applications where current inputs need to be interpreted in the light of previous experience and knowledge.

General Classification Object Recognition +1

What am I Searching for: Zero-shot Target Identity Inference in Visual Search

1 code implementation31 Jul 2018 Mengmi Zhang, Gabriel Kreiman

Using those error fixations, we developed a model (InferNet) to infer what the target was.

Egocentric Spatial Memory

1 code implementation31 Jul 2018 Mengmi Zhang, Keng Teck Ma, Shih-Cheng Yen, Joo Hwee Lim, Qi Zhao, Jiashi Feng

Egocentric spatial memory (ESM) defines a memory system with encoding, storing, recognizing and recalling the spatial information about the environment from an egocentric perspective.

Feature Engineering

Egocentric Spatial Memory Network

no code implementations ICLR 2018 Mengmi Zhang, Keng Teck Ma, Joo Hwee Lim, Shih-Cheng Yen, Qi Zhao, Jiashi Feng

During the exploration, our proposed ESM network model updates belief of the global map based on local observations using a recurrent neural network.

Navigate Simultaneous Localization and Mapping

Deep Future Gaze: Gaze Anticipation on Egocentric Videos Using Adversarial Networks

1 code implementation CVPR 2017 Mengmi Zhang, Keng Teck Ma, Joo Hwee Lim, Qi Zhao, Jiashi Feng

Through competition with discriminator, the generator progressively improves quality of the future frames and thus anticipates future gaze better.

Gaze Prediction

Cannot find the paper you are looking for? You can Submit a new open access paper.