Keras Deep Computer Vision

This repository contains model definitions, training scripts, and other examples for Keras (Tensorflow backend) implementations for classification, detection, and segmentation (computer vision).

Models

Classification

• [x] LeNet Paper Model
• [x] AlexNet Paper Model
• [x] VGG16 and VGG19 Paper Model
• [ ] ResNet Paper
• [ ] YOLO9000 Paper
• [ ] DenseNet Paper

Detection

• [ ] Faster RCNN Paper
• [ ] SSD Paper
• [ ] YOLOv2 Paper
• [ ] R-FCN Paper

Segmentation

• [ ] FCN8 Paper
• [ ] SegNet Paper
• [ ] U-Net Paper
• [ ] E-Net Paper
• [ ] ResNetFCN Paper
• [ ] PSPNet Paper
• [ ] Mask RCNN Paper

Datasets

Classification

• [x] MNIST
• [x] CIFAR-10
• [ ] MNIST-Fashion
• [ ] ImageNet
• [ ] Pascal VOC

Detection

• [ ] Pascal VOC
• [ ] LISA Traffic Sign
• [ ] KITTI
• [ ] MSCOCO

Segmentation

• [ ] CamVid
• [ ] Cityscapes
• [ ] Pascal VOC
• [ ] KITTI
• [ ] SYNTHIA
• [ ] GTA-V Segmentation
• [ ] MSCOCO

Prerequisites

For the models in thie repo, Keras and Tensorflow are required. Make sure the latest versions are installed.

After these packages have been installed, a few other packages are required (which can be found in requirements.txt)

pip install -r requirements.txt

Acknowledgments

This repo would like to acknowledge the following pieces of code which played a part in the development of this repository:

• keras_zoo