Project Objective
Save search queries for the index in Kibana on AWS. Make visualizations to view the yearly distribution of the image on the basis of the model of the camera and company of camera and tables to get the frequency of label of the image.
Project Description
Create search queries for image data for the following queries
- Photos over time (taken, uploaded)
- Images by camera
- Images by camera over time
- Images by labels (needs to include nested labels)
- Images by labels over time
and provide details regarding the usage of queries to filter data.
Make visualizations to display the yearly distribution of the image on the basis of the model of the camera and the company of the camera. Also, put a table to show the frequency distribution of the model within the camera company and a table for the count of labels of the images.
Our Solution
Created search queries on Kibana which filters data according to the mentioned queries with the editable empty field value.
Following are the solutions for the respective query
- Photos over time (taken, uploaded)
Query: exifData.exif.exif.DateTimeOriginal >”YYYY-MM-DD” and exifData.exif.exif.DateTimeOriginal <“YYYY-MM-DD”
by replacing “YYYY-MM-DD” with start date and end date, user can set the range of the data that user is interested in
- Images by camera
Query: exifData.LensModel.description : “camera” and exifData.LensModel.value: “camera”
by replacing “camera” with the model name user is interested in, this query will filter the data with the camera model name input by the user
- Images by camera over time
Query: (exifData.LensModel.description : “camera” and exifData.LensModel.value: “camera”) and (exifData.exif.exif.DateTimeOriginal >”YYYY-MM-DD” and exifData.exif.exif.DateTimeOriginal <“YYYY-MM-DD”)
by replacing “camera” with the model name user is interested in and replacing “YYYY-MM-DD” with start date and end date, user can retrieve the data with a specific camera model in the input date range from this query
- Images by labels (needs to include nested labels)
Query: (Labels.Name: “Person1″ or Labels.Name:”Human”) and (Labels.Parents.Name.keyword: “Person2”)
by replacing “Person1″/”Human” with the label name user is interested in and replacing “Person2” with the label name of parent user is interested in, this query will filter the data with the label name and label name of parent input by the user
- Images by labels over time
Query: (Labels.Name: “Person1″ or Labels.Name:”Human”) and (Labels.Parents.Name.keyword: “Person2″) and (exifData.exif.exif.DateTimeOriginal >”YYYY-MM-DD” and exifData.exif.exif.DateTimeOriginal <“YYYY-MM-DD”)
by replacing “Person1″/”Human” with the label name user is interested in and replacing “Person2” with the label name of parent user is interested in and replacing “YYYY-MM-DD” with start date and end date, user can retrieve the data with a specific or more label name and label name of parent in the input date range from this query
Created 2 visualizations showing yearly distribution of the image one on the basis of model of camera and other for company of camera. Made a count table to show the frequency of models with respect to company of camera. Made frequency table of Label names as well.
Project Deliverables
2 visualizations
5 search queries
2 tables
Tools used
Kibana & Elasticsearch
Language/techniques used
Lucene language is used on Kibana to make search queries
Skills used
Visualization and analytical skills were used
Databases used
Image label dataset is used
Web Cloud Servers used
Amazon Web Services
What are the technical Challenges Faced during Project Execution
Too many fields made it a bit ambiguous firstly while understanding the existing fields
How the Technical Challenges were Solved
The client helped in explaining a few terms and fields, then it fields were quite clear while exploring data
