(Version 0.1, Last updated :2018.07.02)

This is mxnet implementation of the YOLO:Real-Time Object Detection. YOLO is an unified framework for object detection with a single network.

It has been originally introduced in this research article.

This repository contains a MxNet implementation of a MobileNets_V2-based YOLO networks.

For details with Google's MobileNets, please read the following papers:

• [v1] MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• [v2] Inverted Residuals and Linear Bottlenecks: Mobile Networks for Classification, Detection and Segmentation

See: https://github.com/yuantangliang/MobileNet-v2-Mxnet

The top-1/5 accuracy rates by using single center crop (crop size: 224x224, image size: 256xN):

For testing model purposes, we’ll train our model to detect Pikachu in the wild. We use a synthetic toy dataset by rendering images from open-sourced 3D Pikachu models.

For more detail. Please see：

• https://gluon.mxnet.io/chapter08_computer-vision/object-detection.html.
• http://zh.gluon.ai/chapter_computer-vision/pikachu.html.

The dataset consists of 1088 pikachus with random pose/scale/position in random background images. The exact locations are recorded as ground-truth for training and validation.