https://github.com/vbhavank/challenge_question2_catdog_vasu
This is a project for a CatvsDog classifier using a very limited number of samples
https://github.com/vbhavank/challenge_question2_catdog_vasu
catvsdog-classifier computer-vision deep-learning machine-learning python tensorflow
Last synced: about 2 months ago
JSON representation
This is a project for a CatvsDog classifier using a very limited number of samples
- Host: GitHub
- URL: https://github.com/vbhavank/challenge_question2_catdog_vasu
- Owner: vbhavank
- Created: 2018-01-16T05:40:36.000Z (over 8 years ago)
- Default Branch: master
- Last Pushed: 2018-01-30T09:56:45.000Z (over 8 years ago)
- Last Synced: 2025-03-05T01:45:14.058Z (over 1 year ago)
- Topics: catvsdog-classifier, computer-vision, deep-learning, machine-learning, python, tensorflow
- Language: Python
- Homepage:
- Size: 265 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Training a CNN with just 20 images
Author:
Bhavan Vasu
Graduate Research Assistant,
Real-time and computer vision lab,
Rochester Institute of Technology,
New York-14623
bxv7657@rit.edu
Requirements :
1) Numpy
2) matplotlib
3) tflearn
4) Opencv2
Add the training and testing samples into two folders in the current directory,named 'train' and 'test' respectively.
Alexnet was chosen for the implementation of a CatvsDog classifier using Tensorflow and python.(Check network graph for layer information)
The network is trained on 18 images and validated on 2 images during training.
The network is trained for just 12 epochs with a batch size of 16 with a learning rate of 1e-5.
The network manages to achieve a test accuracy of about 50-70%, measured from visual inspection of the 20 unknown test images.
Run the 'catd.py' file for the CatvsDog classifier.
To test and train:
$ python catd.py
To run tensorboard for network visualization
$tensorboard --logdir="logs"