• Publications
  • Influence
Soft-NMS — Improving Object Detection with One Line of Code
TLDR
We propose Soft-NMS, an algorithm which decays the detection scores of all other objects as a continuous function of their overlap with M. Expand
  • 499
  • 97
  • PDF
An Analysis of Scale Invariance in Object Detection - SNIP
  • B. Singh, L. Davis
  • Computer Science
  • IEEE/CVF Conference on Computer Vision and…
  • 22 November 2017
TLDR
We present a novel training scheme called Scale Normalization for Image Pyramids (SNIP) which selectively back-propagates the gradients of object instances of different sizes as a function of the image scale. Expand
  • 273
  • 30
  • PDF
A Multi-stream Bi-directional Recurrent Neural Network for Fine-Grained Action Detection
TLDR
We present a multi-stream bi-directional recurrent neural network for fine-grained action detection. Expand
  • 265
  • 25
  • PDF
Training Neural Networks Without Gradients: A Scalable ADMM Approach
TLDR
This paper explores an unconventional training method that uses alternating direction methods and Bregman iteration to train networks without gradient descent steps. Expand
  • 161
  • 25
  • PDF
SNIPER: Efficient Multi-Scale Training
TLDR
We present SNIPER, an algorithm for performing efficient multi-scale training in instance level visual recognition tasks, which adaptively samples chips from multiple scales in an image pyramid. Expand
  • 166
  • 23
  • PDF
Temporal Context Network for Activity Localization in Videos
TLDR
We present a Temporal Context Network (TCN) for precise temporal localization of human activities. Expand
  • 149
  • 9
  • PDF
R-FCN-3000 at 30fps: Decoupling Detection and Classification
TLDR
We propose a modular approach to the largescale object detection problem that outperforms YOLO9000 by 18% while processing 30 images per second. Expand
  • 46
  • 7
  • PDF
Deception Detection in Videos
TLDR
We present a system for covert automated deception detection in real-life courtroom trial videos. Expand
  • 35
  • 7
  • PDF
Selecting Relevant Web Trained Concepts for Automated Event Retrieval
TLDR
We propose an event retrieval algorithm that constructs pairs of automatically discovered concepts and then prunes those concepts that are unlikely to be helpful for retrieval. Expand
  • 32
  • 6
  • PDF
Fast-At: Fast Automatic Thumbnail Generation Using Deep Neural Networks
TLDR
We propose Fast-AT, a deep learning based approach for thumbnail generation that addresses the problem directly, in an end-to-end learning framework. Expand
  • 24
  • 5
  • PDF