CDIoU and CDIoU loss is like a convenient plug-in that can be used in multiple models. CDIoU and CDIoU loss have different excellent performances in several models such as Faster R-CNN, YOLOv4, RetinaNet and . There is a maximum AP improvement of 1.9% and an average AP of 0.8% improvement on MS COCO dataset, compared to traditional evaluation-feedback modules. Here we just use as an example to illustrate the code.

Overview

CDIoU-CDIoUloss

CDIoU and CDIoU loss is like a convenient plug-in that can be used in multiple models. CDIoU and CDIoU loss have different excellent performances in several models such as Faster R-CNN, YOLOv4, RetinaNet and . There is a maximum AP improvement of 1.9% and an average AP of 0.8% improvement on MS COCO dataset, compared to traditional evaluation-feedback modules. Here we just use as an example to illustrate the code.

Control Distance IoU and Control Distance IoU Loss Function

by Chen Dong, Miao Duoqian

Introduction

Numerous improvements for feedback mechanisms have contributed to the great progress in object detection. In* this paper, we first present an evaluation-feedback module, which is proposed to consist of evaluation system and feedback mechanism. Then we analyze and summarize the disadvantages and improvements of traditional evaluation-feedback module. Finally, we focus on both the evaluation system and the feedback mechanism, and propose Control Distance IoU and Control Distance IoU loss function (or CDIoU and CDIoU loss for short) without increasing parameters or FLOPs in models, which show different significant enhancements on several classical and emerging models. Some experiments and comparative tests show that coordinated evaluation-feedback module can effectively improve model performance. CDIoU and CDIoU loss have different excellent performances in several models such as Faster R-CNN, YOLOv4, RetinaNet and ATSS. There is a maximum AP improvement of 1.9% and an average AP of 0.8% improvement on MS COCO dataset, compared to traditional evaluation-feedback modules.

There are some potential defects in the current mainstream target detection

  • It relies too much on the deepening of the backbone to extract features, so as to improve the accuracy of target detection;

  • The deepening of neural network, especially the deepening of backbone and neck, results in huge parameters and flops of the model;

  • Compared with the evaluation system (IoUs, the common ones are IoU and GIoU). At present, some new model optimization focuses more on the feedback mechanism (IoU losses), such as IoU loss, smooth loss, GIoU loss,CIoU loss, DIoU loss.

We propose Control Distance IoU and Control Distance IoU Loss Function (CDIoU and CDIoU loss for short).

Analysis of traditional IoUs and loss functions

  • Analysis of traditional IoUs

  • IoU: Smooth L1 Loss and IoU Loss

  • GIoU and GIoU Loss

  • DIoU loss and CIoU Loss

For more information, see Control Distance IoU and Control Distance IoU Loss Function for Better Bounding Box Regression

Installation

CDIoU and CDIoU loss is like a convenient plug-in that can be used in multiple models. CDIoU and CDIoU loss have different excellent performances in several models such as Faster R-CNN, YOLOv4, RetinaNet and ATSS. There is a maximum AP improvement of 1.9% and an average AP of 0.8% improvement on MS COCO dataset, compared to traditional evaluation-feedback modules. Here we just use ATSS as an example to illustrate the code.

These models use different frameworks, and some even have versions, so no code is provided in this article.


This ATSS implementation is based on FCOS and maskrcnn-benchmark and the installation is the same as them. Please check INSTALL.md for installation instructions.

ATSS bridges the gap between anchor-based and anchor-free detection via adaptive training sample selection. Comparison tests on ATSS exclude the essential interference between anchor-based and anchor-free detection. In these tests, the interference of positive and negative sample generation is eliminated, which give tests based on ATSS more representativeness.

CDIoU and CDIoU loss functions

CDIoU

Turning

For more information, see Control Distance IoU and Control Distance IoU Loss Function for Better Bounding Box Regression

Experiments

In order to verify the effectiveness of CDIoU and CDIoU loss in object detection, experiments are designed and applied to numerous models in this paper. These models encompass existing classical models and emerging models, reflecting certain robustness and wide adaptability.

image-20210315210941666

image-20210315214303929

For more information, see Control Distance IoU and Control Distance IoU Loss Function for Better Bounding Box Regression

Models

For your convenience, we provide the following trained models. All models are trained with 16 images in a mini-batch and frozen batch normalization (i.e., consistent with models in FCOS and maskrcnn_benchmark).

Model Multi-scale evaluation system feedback mechanism AP (val) AP (test-dev) pth
ATSS R 50 FPN 1x + CDIoU & loss NO CDIoU CDIoU loss 39.5 39.4 ATSS R 50 FPN 1x + CDIoU & loss
ATSS dcnv2 R 50 FPN 1x + CDIoU & loss NO CDIoU CDIoU loss 43.1 43.1 ATSS dcnv2 R 50 FPN 1x + CDIoU & loss
ATSS dcnv2 R 101 FPN 2x + CDIoU & loss NO CDIoU CDIoU loss 46.3 46.4 ATSS dcnv2 R 101 FPN 2x + CDIoU & loss
ATSS X 101 32x8d FPN 2x + CDIoU & loss NO CDIoU CDIoU loss 45.1 45.2 ATSS X 101 32x8d FPN 2x + CDIoU & loss
ATSS dcnv2 X 101 32x8d FPN 2x + CDIoU & loss NO CDIoU CDIoU loss 48.1 47.9 ATSS dcnv2 X 101 32x8d FPN 2x + CDIoU & loss
ATSS dcnv2 X 101 32x8d FPN 2x(MS) + CDIoU & loss YES CDIoU CDIoU loss 50.9 50.7 ATSS dcnv2 X 101 32x8d FPN 2x(MS) + CDIoU & loss

[1] The testing time is taken from FCOS, because our method only redefines positive and negative training samples without incurring any additional overhead. [2] 1x and 2x mean the model is trained for 90K and 180K iterations, respectively. [3] All results are obtained with a single model and without any test time data augmentation such as multi-scale, flipping and etc.. [4] dcnv2 denotes deformable convolutional networks v2. Note that for ResNet based models, we apply deformable convolutions from stage c3 to c5 in backbones. For ResNeXt based models, only stage c4 and c5 use deformable convolutions. All models use deformable convolutions in the last layer of detector towers. [5] The model ATSS_dcnv2_X_101_64x4d_FPN_2x with multi-scale testing achieves 50.7% in AP on COCO test-dev. Please use TEST.BBOX_AUG.ENABLED True to enable multi-scale testing.

MSCOCO test-dev

image-20210315222453294

Tips to improve performances

  • Floating learning rate

It is a consensus that the learning rate decreases as the iterative process in the experiment. Further, this paper proposes to check the loss every K iterations and increase the learning rate slightly, if the loss function does not decrease continuously. In this way, the learning rate will decrease and float appropriately at regular intervals to promote the decrease of the loss function.

  • Automatic GT clustering analysis

It is well known that AP can be effectively improved by performing cluster analysis on GT in the original dataset. We adjust anchor sizes and aspect ratios parameters based on the results of this cluster analysis. However, we do not know the number of clusters through the current approach. The main solution is to keep trying the number of clusters N , and then judge by the final result AP. Obviously, this exhaustive method takes a lot of time.

Contributing to the project

Any pull requests or issues are welcome.

CItations

Please cite our paper in your publications if it helps your research: And is not true!!

<_>This reference stuff is for fun only!!!!!!!!!

@inproceedings{chen2021CDIoU,
  title     =  {Control Distance IoU and Control Distance IoU Loss Function for Better Bounding Box Regression},
  author    =  {Chendong, Miaoduoqian.},
  booktitle =  {ICCV},
  year      =  {2021}
}
Owner
Alan D Chen
UJN : bachelor : MATH &CS | TONGJI : PhD : CV
Alan D Chen
Receive GitHub webhook events and send to Telegram chats with AIOHTTP through Telegram Bot API

GitHub Webhook to Telegram Receive GitHub webhook events and send to Telegram chats with AIOHTTP through Telegram Bot API What this project do is very

Dash Eclipse 33 Jan 03, 2023
Trading Strategies (~50%) developed by GreenT on QuantConnect platform over the autumn quarter

Trading Strategies ~50% of codes from the Applied Financial Technology Course. Contributors: Claire W. Derrick T. Frank L. Utkarsh T. Course Leads: Dy

Utkarsh 2 Feb 07, 2022
Simple Telegram Bot to Download and Upload Files From Mega.nz

Mega.nz-Bot Simple Telegram Bot to Download Files From Mega.nz and Upload It to Telegram Features All Mega.nz File Links supported No login required A

I'm Not A Bot #Left_TG 245 Jan 01, 2023
Jupyter notebooks and AWS CloudFormation template to show how Hudi, Iceberg, and Delta Lake work

Modern Data Lake Storage Layers This repository contains supporting assets for my research in modern Data Lake storage layers like Apache Hudi, Apache

Damon P. Cortesi 25 Oct 31, 2022
Discord bot that automatically fills out health screenings

Auto Covid Bot Automatically fill out the NYC DOE health screening form by registering with a discord bot School code can be found on https://schoolse

Cleo 2 Jul 29, 2022
5 Oct 19, 2022
historical code from reddit.com

This repository is archived. This repository is archived and will not receive any updates or accept issues or pull requests. To report bugs in reddit.

The Reddit Archives 16.3k Dec 31, 2022
A Discord webhook spammer made in Python.

A Python made Discord webhook spammer usually used for token loggers to spam them/delete them original by cattyn I only made it so u can change the avatar to whatever u want instead of it being hardc

notperry1234567890 15 Dec 15, 2021
A delivery protection and notification system

DeliveryProtect This project builds a delivery protection and notification system, based on integration of Arduino Uno and Raspberry Pi 4. The codes a

2 Dec 13, 2021
A Twitch bot to provide information from the WebNowPlayingCompanion extension

WebNowPlayingTwitch A Twitch bot made using TwitchIO which displays information obtained from the WebNowPlaying Compaion browser extension. Image is o

NiceAesth 1 Mar 21, 2022
Public release of Telepathy, an OSINT toolkit for investigating Telegram groups. Enhanced features and improvements will be added over time.

Telepathy Welcome to Telepathy, an OSINT toolkit for scraping Telegram data to help investigate shady goings on. Currently, the tool is limited to scr

Jordan Wildon 484 Jan 01, 2023
Unofficial API wrapper for seedr.cc

Seedr API Unofficial API wrapper for seedr.cc Inspired by theabbie's seedr-api Powered by @harp_tech (Telegram) How to use You can install lib via git

Anjana Madu 49 Oct 24, 2022
A Python script that wraps the gitleaks tool to enable scanning of multiple repositories in parallel

mpgitleaks A Python script that wraps the gitleaks tool to enable scanning of multiple repositories in parallel. The motivation behind writing this sc

Emilio Reyes 7 Dec 29, 2022
Modern, privacy-friendly, and detailed web analytics that works without cookies or JS.

Modern, privacy-friendly, and cookie-free web analytics. Getting started » Screenshots • Features • Office Hours Motivation There are a lot of web ana

R. Miles McCain 2.1k Jan 03, 2023
Automatically changes your discord status

Automatically changes your discord status, Be careful as this may get you rate limited and banned

octo 5 Sep 20, 2022
It's My Bot, For my group in telegram :)

Get Start USage This robot is written in Python language for devdood Group in Telegram ... You can easily edit and use this source Edit and Run You ne

Mohsen farzadmanesh 7 Sep 24, 2022
Event-driven-model-serving - Unified API of Apache Kafka and Google PubSub

event-driven-model-serving Unified API of Apache Kafka and Google PubSub 1. Proj

Danny Toeun Kim 4 Sep 23, 2022
A program that automates the boring parts of completing the Daily accounting spreadsheet at Taos Ski Valley

TSV_Daily_App A program that automates the boring parts of completing the Daily accounting spreadsheet at my old job. To see how it works you will nee

Devin Beck 2 Jan 01, 2022
Access LeetCode problems via id

LCid - access LeetCode problems via id Introduction As a world's leading online programming learning platform, LeetCode is quite popular among program

bunnyxt 14 Oct 08, 2022
Stackoverflow Telegram Bot With Python

Template for Telegram Bot Template to create a telegram bot in python. How to Run Set your telegram bot token as environment variable TELEGRAM_BOT_TOK

PyTopia 10 Mar 07, 2022