CMATERdb | Easy to use CMATERdb datasets converted in NumPy format | Machine Learning library
kandi X-RAY | CMATERdb Summary
kandi X-RAY | CMATERdb Summary
CMATERdb is the pattern recognition database repository created at the ‘Center for Microprocessor Applications for Training Education and Research’ (CMATER) research laboratory, Jadavpur University, Kolkata 700032, INDIA. This database is free for all non-commercial uses. Please acknowledge CMATER explicitly, whenever you use this database for academic and research purposes. For using some databases, one must also cite relevant research publications, mentioned in this website. Official Dataset Repository: Link Shifted Repository as per Google Code Archive (Not Live): Link.
Support
Quality
Security
License
Reuse
Top functions reviewed by kandi - BETA
- Save the given numeral_type .
- Load numeral images .
CMATERdb Key Features
CMATERdb Examples and Code Snippets
Community Discussions
Trending Discussions on CMATERdb
QUESTION
I want to predict the digit is 5 or not from 0 to 9.I have used cmaterdb dataset.
For this task I have changed all the digit label 0 except 5 in the training and test dataset
...ANSWER
Answered 2020-May-29 at 20:09I can guess, you have an accuracy of about 90%, but your model almost always predicts 0 ( not 5), right?
If it is the case, the reason can be the imbalanced classes. Your original dataset, I guess is balanced, like we have almost N 0s, almost N 1s, ... and almost N 9s. What you did, you kept the almost N 5s, and labeled it to 1, and turn the rest of almost 9*N examples to label 0. It means now you have 10% of data labeled 1( which are 5s) and 90% of data labeled 0. This is an imbalance case, and it is very normal you get a good accuracy of around 90%, but a poor prediction for the minority class. For imbalanced cases, accuracy is not a good metric or maybe it is not enough. Try to trace Precision,recall, and F1 as well.
my suggestion, sample from the non 5 classes ( 1/10 from each ) and keep all the 5s. You should get better predictions.
QUESTION
How do I format a dataset for training in Python?
I have 3000 grayscale BMP images of some handwritten digits (just like MNIST). Now I want to train my model with this dataset (I am using the Keras library) and I am using a convolution neural network for training.
I am using this code to convert one of the images into array
...ANSWER
Answered 2017-Jul-06 at 18:03You can use any library that loads image files into arrays, such as Pillow.
Read Pillow's documentation to learn how to load an image file into an array.
Then, you should usually scale the array into values between 0 and 1. Usually, you just divide the image array by 255 (because they are RGB values between 0 and 255).
Be sure to end up with an array shaped like this: (3000, width, heigth, channels), where channels is usually 3 (Red, green, blue).
Community Discussions, Code Snippets contain sources that include Stack Exchange Network
Vulnerabilities
No vulnerabilities reported
Install CMATERdb
You can use CMATERdb like any standard Python library. You will need to make sure that you have a development environment consisting of a Python distribution including header files, a compiler, pip, and git installed. Make sure that your pip, setuptools, and wheel are up to date. When using pip it is generally recommended to install packages in a virtual environment to avoid changes to the system.
Support
Reuse Trending Solutions
Find, review, and download reusable Libraries, Code Snippets, Cloud APIs from over 650 million Knowledge Items
Find more librariesStay Updated
Subscribe to our newsletter for trending solutions and developer bootcamps
Share this Page