Aircraft, as a special strategic target, has high value in both civil and military use, and it is especially important to achieve intelligent scientific aircraft target detection compared to traditional methods. There...
详细信息
Inefficient garbage collection not only leads to overflowing bins and unpleasant but also poses significant environmental and health risks. This paper proposes an effective garbage monitoring system utilizing a GSM mo...
详细信息
Healthcare systems all over the world are strained as the COVID-19 pandemic39;s spread becomes more widespread. The only realistic strategy to avoid asymptomatic transmission is to monitor social distance, as there ...
详细信息
India, where a third of the world39;s blind people reside, has about 12 million blind people, especially in comparison to a total of 39 million worldwide, according to the National Programme for control of Blindness...
详细信息
This project introduces a comprehensive home automation system leveraging Wi-Fi connectivity and ESP32 micro controllers, aimed at enhancing modern living through seamless automation and user-centric design. The backg...
详细信息
As the main carrier of Metaverse VR technology, the head-mounted display system plays a vital role in leading this technological trend. This paper implements a head-mounted display system based on Metaverse VR technol...
详细信息
The proceedings contain 52 papers. The special focus in this conference is on Trends in Electronics and Health Informatics. The topics include: An Effective Combination of Deep and Machine Learning Models for Monkeypo...
ISBN:
(纸本)9789819739363
The proceedings contain 52 papers. The special focus in this conference is on Trends in Electronics and Health Informatics. The topics include: An Effective Combination of Deep and Machine Learning Models for Monkeypox Detection from Dermatographic Image;a Time-Efficient and Effective Image Contrast Enhancement Technique Using Fuzzification and Defuzzification;An Ensemble Machine Learning-Based Approach for Detecting Malicious Websites Using URL Features;The Multi-class Paradigm: How Transformers Are Reshaping Language Analysis in NLP;deep Learning Precision Farming: Identification of Bangladeshi-Grown Fruits Using Transfer Learning-Based Detection;Deep Learning Solutions for Detecting Bangla Fake News: A CNN-Based Approach;a Two-Stage Stacking Ensemble Learning for Employee Attrition Prediction;ensemble Learning Approaches for Alzheimer’s Disease Classification in Brain Imaging Data;pseudo-Knighted Cocktail Shaker Sort;sentiment Analysis in Twitter Data Using Machine Learning-Based Approach;road Object Detection for Visually Impaired People in Bangladesh;newBreeze: A Comprehensive Solution to a Beginner-Friendly Arch Linux Distribution with Zen Kernel;deep Ensemble Learning Approach for Multimodal Emotion Recognition;Tri Focus Net: A CNN-Based Model with Integrated Attention Modules for Pest and Insect Detection in Agriculture;detection and Classification of Spam Email: A Machine Learning-Based Experimental Analysis;predictive Modeling and Early Detection of White Spot Disease in Shrimp Farming Using Machine Learning: A Case Study in Bangladesh;bangla License Plate Detection and Recognition Approach Based on computervision for Authentic Vehicle Identification;feature Techniques with a Custom Convolutional Model for Breast Tumor Surveillance in Mammograms;An AI-Based Clinical Recommendation System Using Ensemble-Based Soft Voting Classifier;machine Learning-Based Approach to Predict Heart Diseases Using Fused Dataset;an Optimal Feature Selection-Based Approach to P
We present an approach to efficiently and effectively adapt a masked image modeling (MIM) pre-trained vanilla vision Transformer (ViT) for object detection, which is based on our two novel observations: (i) A MIM pre-...
ISBN:
(纸本)9798350307184
We present an approach to efficiently and effectively adapt a masked image modeling (MIM) pre-trained vanilla vision Transformer (ViT) for object detection, which is based on our two novel observations: (i) A MIM pre-trained vanilla ViT encoder can work surprisingly well in the challenging object-level recognition scenario even with randomly sampled partial observations, e.g., only 25% similar to 50% of the input embeddings. (ii) In order to construct multi-scale representations for object detection from single-scale ViT, a randomly initialized compact convolutional stem supplants the pre-trained patchify stem, and its intermediate features can naturally serve as the higher resolution inputs of a feature pyramid network without further upsampling or other manipulations. While the pre-trained ViT is only regarded as the 3rd-stage of our detector's backbone instead of the whole feature extractor. This naturally results in a ConvNet-ViT hybrid architecture. The proposed detector, named MIMDET, enables a MIM pre-trained vanilla ViT to outperform leading hierarchical architectures such as Swin Transformer, MViTv2 and ConvNeXt on COCO object detection & instance segmentation, and achieves better results compared with the previous best adapted vanilla ViT detector using a more modest fine-tuning recipe while converging 2.8x faster. Code and pre-trained models are available at https://***/hustvl/MIMDet.
The K-nearest neighbor (KNN) algorithm is widely used in navigation, such as traffic management, driverless vehicles, and logistics planning. While it offers powerful instance-based learning, its performance can be in...
详细信息
Environmental sustainability is one of the sustainable development goals and the invention of electric vehicles marks the beginning of alternatives to fuel-based vehicles. Electric vehicles have penetrated the global ...
详细信息
暂无评论