Guess what's on the image?

Let me predict the object class from your image

Drop your Image here
or click to select file



{{ name }} {{ value }}%
{{ error }}

Google Inception V4

About Algorithm

Google Inception architecture is one of the top advances in image recognition performance achieved with relatively low computational cost. Inception-v4 neural network combines traditional network architecture with residual connections, which allowed to significantly accelerate the training and achieve state-of-the-art performance in 2015 ILSVRC challenge.

Inception-v4

Features

Module

TensorFlow

Top-1 Accuracy

80.6%

Top-5 Error

3.08%

Network Type

Deep convolutional network

Activation Function

ReLu

Number of Parameters

43 million

Optimizer

RMSprop (Decay 0.9)

Learning rate

0.045

Inception Code

GitHub Repository

Copyright 2016 The TensorFlow Authors. All Rights Reserved. Inception Creators: Christian Szegedy, Sergey Ioffe, Vincent Vanhoucke, Alex Alemi

Visit other projects

Septellar

blockchain

Best Value Stocks

pdf

PDF Text Extractor

thompson

Thompson Sampling

quant

Quant Properties

cat-dog

Cat Dog Classifier

utf

UTF-8 Converter

bepartisan

Yahoo Woeid Search