Inception V1 Trained on Places365 Data

Identify the scene type of an image

Released in 2016 by the MIT Computer Science and Artificial Intelligence Laboratory as a pre-trained model for the launch of the Places365 dataset (a subset of the Places2 dataset). The model is based on the Inception V1 architecture.

Number of layers: 147 | Parameter count: 6,347,677 | Trained size: 26 MB |

Training Set Information

Places365-Standard dataset, consisting of 1.8 million training images with 365 scene categories.

Performance

This model achieves 53.59% top-1 and 84.01% top-5 accuracy on the Places365-Standard dataset.

Examples

Download Example Notebook

Open in Wolfram Cloud

Resource retrieval

Get the pre-trained net:

In[1]:=

Out[1]=

Basic usage

Find the type of scene of an image:

In[2]:=

(* Evaluate this cell to get the example input *) CloudGet["https://www.wolframcloud.com/obj/a817913b-3304-441e-bd4b-ebb82688778d"]

Out[2]=

Obtain the probabilities of the most likely scenes:

In[3]:=

(* Evaluate this cell to get the example input *) CloudGet["https://www.wolframcloud.com/obj/00abde0b-d496-4d15-8a30-29e8852cda39"]

Out[3]=

Net information

Inspect the number of parameters of all arrays in the net:

In[4]:=

$NetInformation[ NetModel["Inception V1 Trained on Places365 Data"], \ "ArraysElementCounts"]$

Out[4]=

Obtain the total number of parameters:

In[5]:=

$NetInformation[ NetModel["Inception V1 Trained on Places365 Data"], \ "ArraysTotalElementCount"]$

Out[5]=

Obtain the layer type counts:

In[6]:=

$NetInformation[ NetModel["Inception V1 Trained on Places365 Data"], \ "LayerTypeCounts"]$

Out[6]=

Display the summary graphic:

In[7]:=

Out[7]=

Export to MXNet

Export the net into a format that can be opened in MXNet:

In[8]:=

Out[8]=

Export also creates a net.params file containing parameters:

In[9]:=

Out[9]=

Get the size of the parameter file:

In[10]:=

Out[10]=

The size is similar to the byte count of the resource object:

In[11]:=

Out[11]=

Represent the MXNet net as a graph:

In[12]:=

Out[12]=

Construction Notebook

Download Construction Notebook

Open in Wolfram Cloud

Requirements

Wolfram Language 11.1 (March 2017) or above

Resource History

Date Created: 2 February 2017
Latest Update: 21 June 2018

Reference

B. Zhou, A. Khosla, A. Lapedriza, A. Torralba, A. Oliva, "Places: An Image Database for Deep Scene Understanding," arXiv:1610.02055 (2016)
Available from: https://github.com/metalbubble/places365
Rights: Creative Commons Attribution 4.0 International (CC BY 4.0)