paper-with-me

Papers

AQD: Towards Accurate Fully-Quantized Object Detection

2020-07-14 · CVPR 2021 1 · Peng Chen, Jing Liu, Bohan Zhuang, Mingkui Tan, Chunhua Shen

Network quantization allows inference to be conducted using low-precision arithmetic for improved inference efficiency of deep neural networks on edge devices. However, designing aggressively low-bit (e.g., 2-bit) quantization schemes on complex tasks, such as object detection, still remains challenging in terms of severe performance degradation and unverifiable efficiency on common hardware. In this paper, we propose an Accurate Quantized object Detection solution, termed AQD, to fully get rid of floating-point computation. To this end, we target using fixed-point operations in all kinds of layers, including the convolutional layers, normalization layers, and skip connections, allowing the inference to be executed using integer-only arithmetic. To demonstrate the improved latency-vs-accuracy trade-off, we apply the proposed methods on RetinaNet and FCOS. In particular, experimental results on MS-COCO dataset show that our AQD achieves comparable or even better performance compared with the full-precision counterpart under extremely low-bit schemes, which is of great practical value. Source code and models are available at: https://github.com/ziplab/QTool

📄 PDF Abstract BibTeX arXiv:2007.06919

Code (1)

ziplab/qtool 공식 구현 pytorch

Tasks

Image ClassificationObjectobject-detectionObject DetectionQuantization

Methods 이 논문이 사용한 방법론

Non Maximum Suppression Non Maximum Suppression is a computer vision method that selects a single entity out of many overlapping entities (for example bounding boxes in object detection). The…
FCOS FCOS is an anchor-box free, proposal free, single-stage object detection model. By eliminating the predefined set of anchor boxes, FCOS avoids computation related to anchor…
Convolution A convolution is a type of matrix operation, consisting of a kernel, a small matrix of weights, that slides over input data performing element-wise multiplication with the…
Focal Loss A Focal Loss function addresses class imbalance during training in tasks like object detection. Focal loss applies a modulating term to the cross entropy loss in order to…
1x1 Convolution A 1 x 1 Convolution is a convolution with some special properties in that it can be used for dimensionality reduction,…
FPN 설명 없음
RetinaNet RetinaNet is a one-stage object detection model that utilizes a focal loss function to address class imbalance during training.…
Batch Normalization 설명 없음

Similar Papers 제목 키워드 기반

Fully Quantized Network for Object Detection

2019-06-01 · CVPR 2019 6 · Rundong Li, Yan Wang, Feng Liang, Hongwei Qin 외

Efficient neural network inference is important in a number of practical domains, such as deployment in mobile settings. An effective method for increasing inference efficiency is to use low bitwidth arithmetic, which c…

Efficient Neural NetworkObjectobject-detectionObject Detection+1

AttentionNet: Aggregating Weak Directions for Accurate Object Detection

2015-06-25 · ICCV 2015 12 · Donggeun Yoo, Sunggyun Park, Joon-Young Lee, Anthony S. Paek 외

We present a novel detection method using a deep convolutional neural network (CNN), named AttentionNet. We cast an object detection problem as an iterative classification problem, which is the most suitable form of a CN…

General ClassificationHuman DetectionObjectobject-detection+2

Reg-PTQ: Regression-specialized Post-training Quantization for Fully Quantized Object Detector

2024-01-01 · CVPR 2024 1 · Yifu Ding, Weilun Feng, Chuyan Chen, Jinyang Guo 외

Although deep learning based object detection is of great significance for various applications it faces challenges when deployed on edge devices due to the computation and energy limitations. Post-training quantizat…

Objectobject-detectionObject DetectionQuantization+1

LCDet: Low-Complexity Fully-Convolutional Neural Networks for Object Detection in Embedded Systems

2017-05-16 · Subarna Tripathi, Gokce Dane, Byeongkeun Kang, Vasudev Bhaskaran 외

Deep convolutional Neural Networks (CNN) are the state-of-the-art performers for object detection task. It is well known that object detection requires more computation and memory than image classification. Thus the cons…

Face Detectionimage-classificationImage ClassificationObject+3

Reducing the Side-Effects of Oscillations in Training of Quantized YOLO Networks

2023-11-09 · Kartik Gupta, Akshay Asthana

Quantized networks use less computational and memory resources and are suitable for deployment on edge devices. While quantization-aware training QAT is the well-studied approach to quantize the networks at low precision…

object-detectionObject DetectionQuantizationSemantic Segmentation