no code implementations • 11 Aug 2023 • Yen Nhi Truong Vu, Dan Guo, Ahmed Taha, Jason Su, Thomas Paul Matthews
Deep-learning-based object detection methods show promise for improving screening mammography, but high rates of false positives can hinder their effectiveness in clinical practice.
no code implementations • 29 Mar 2023 • Trevor Tsue, Brent Mombourquette, Ahmed Taha, Thomas Paul Matthews, Yen Nhi Truong Vu, Jason Su
The original model trained on both datasets achieved a 0. 945 AUC on the combined US+UK dataset but paradoxically only 0. 838 and 0. 892 on the US and UK datasets, respectively.
1 code implementation • 11 Aug 2022 • Ahmed Taha, Yen Nhi Truong Vu, Brent Mombourquette, Thomas Paul Matthews, Jason Su, Sadanand Singh
In this paper, we tackle this complexity by leveraging a linear self-attention approximation.
1 code implementation • CVPR 2021 • Ahmed Taha, Abhinav Shrivastava, Larry Davis
We evaluate KE using relatively small datasets (e. g., CUB-200) and randomly initialized deep networks.
1 code implementation • 4 Mar 2021 • Ahmed Taha, Alex Hanson, Abhinav Shrivastava, Larry Davis
The SVMax regularizer supports both supervised and unsupervised learning.
2 code implementations • ECCV 2020 • Ahmed Taha, Xitong Yang, Abhinav Shrivastava, Larry Davis
Compared to classification networks, attention visualization for retrieval networks is hardly studied.
no code implementations • 7 Feb 2019 • Ahmed Taha, Yi-Ting Chen, Teruhisa Misu, Abhinav Shrivastava, Larry Davis
We introduce an unsupervised formulation to estimate heteroscedastic uncertainty in retrieval systems.
1 code implementation • 24 Jan 2019 • Ahmed Taha, Yi-Ting Chen, Teruhisa Misu, Abhinav Shrivastava, Larry Davis
We employ triplet loss as a feature embedding regularizer to boost classification performance.
no code implementations • 23 Jan 2019 • Ahmed Taha, Yi-Ting Chen, Xitong Yang, Teruhisa Misu, Larry Davis
We cast visual retrieval as a regression problem by posing triplet loss as a regression loss.
no code implementations • 18 Jun 2018 • Ahmed Taha, Pechin Lo, Junning Li, Tao Zhao
We propose a convolution neural network, called Kid-Net, along with a training schema to segment kidney vessels: artery, vein and collecting system.
no code implementations • 16 Jun 2018 • Ahmed Taha, Moustafa Meshry, Xitong Yang, Yi-Ting Chen, Larry Davis
The self-supervised pre-trained weights effectiveness is validated on the action recognition task.
1 code implementation • 23 Dec 2017 • Rohan Chandra, Sachin Grover, Kyungjun Lee, Moustafa Meshry, Ahmed Taha
A novel loss function, FLTBNK, is used for training the texture synthesizer.
no code implementations • 3 Feb 2017 • Ahmed Taha, Marwan Torki
In our experiments, we evaluate our approach using both human scribble and "robot user" annotations to guide the foreground/background segmentation.