Y

Yolos Small Balloon

Developed by zoheb
YOLOS is an object detection model using Vision Transformer (ViT) architecture, trained with DETR loss and fine-tuned on COCO and Matterport Balloon datasets.
Downloads 101
Release Time : 10/16/2022

Model Overview

This model employs Transformer architecture for object detection, trained with bipartite matching loss, supports PyTorch framework, and is suitable for small-scale object detection tasks.

Model Features

Transformer architecture
Adopts Vision Transformer architecture, processing images as sequential data for object detection.
Bipartite matching loss
Uses Hungarian matching algorithm to establish optimal mapping between predictions and annotations, optimizing the model through cross-entropy and bounding box loss.
Small-scale dataset fine-tuning
Successfully fine-tuned on the Matterport Balloon dataset (only 74 images), demonstrating adaptability to small datasets.

Model Capabilities

Object detection
Bounding box prediction
Small-scale data adaptation

Use Cases

Computer vision
Balloon detection
Detects balloon objects in images and marks their locations
Achieved 26.9 AP on Matterport Balloon validation set
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
Š 2025AIbase