amirunpri2018 / deep-learning-on-custom-images Goto Github PK

Domain             : Computer Vision, Machine Learning
Sub-Domain         : Deep Learning, Image Recognition
Techniques         : Deep Convolutional Neural Network, Data Augumentation
Application        : Image Recognition, Image Classification, Pets Imaging

1. Pets classification using a Deep Convolutional Neural Network by creating a model with 24994 images of cats and dogs (811.5MB).
2. For retraining removed output layers, freezed first few layers and fine-tuned model for two new label classes (Pneumonia and Normal).
3. With Custom Deep Convolutional Neural Network attained testing accuracy 99.50% and loss 0.61.

GitHub Link      : Deep Learning on Custom Images(GitHub)
Linkedin         : Antonio Esteves

Dataset Name     : Cats and Dogs (Pets)
Dataset Link     : Cats and Dogs Dataset
Medium Post      : Agrupando conceitos e classificando imagens com Deep Learning

Dataset Summary
Dataset Name       : Cats and Dogs
Number of Classes            : 2
Images Size                  : Total      : 24994 (811.5 Gigabyte (MB))
                               Train      : 18743 (675.3 Megabyte (MB))
                               Test       : 6251 (224.6 Gigabyte (MB))


Model Params
Machine Learning Library     : Keras
Base Model                   : Custom Deep Convolutional Neural Network
Otimizadores                 : Adam
Função de Perda              : categorical_crossentropy

Deep Convolutional Neural Network: 
Parâmetros de Treino
Batch Size                   : 64
Número of Épocas             : 100
Tempo de Treino              : 1 Hours

Output (Prediction/ Recognition / Classification Metrics)
Test
Accuracy                     : 99.5%
Loss                         : 0.61

Languages                    : Python
Libraries                    : Keras, TensorFlow, Inception, ImageNet

Duration                      : February 2020 - Current
Current Version               : v1.0.0.0
Last Update                   : 27.02.2020