YAD2K: Yet Another Darknet 2 Keras

来源：互联网发布：淘宝每天的成交额编辑：程序博客网时间：2024/04/30 04:58

网址：https://github.com/allanzelener/YAD2K

Welcome to YAD2K

You only look once, but you reimplement neural nets over and over again.

YAD2K is a 90% Keras/10% Tensorflow implementation of YOLO_v2.

Original paper: YOLO9000: Better, Faster, Stronger by Joseph Redmond and Ali Farhadi.

YOLO_v2 COCO model with test_yolo defaults

Requirements

Keras
Tensorflow
Numpy
h5py (For Keras model serialization.)
Pillow (For rendering test results.)
Python 3
pydot-ng (Optional for plotting model.)

Installation

git clone https://github.com/allanzelener/yad2k.gitcd yad2k# [Option 1] To replicate the conda environment:conda env create -f environment.ymlsource activate yad2k# [Option 2] Install everything globaly.pip install numpy h5py pillowpip install tensorflow-gpu  # CPU-only: conda install -c conda-forge tensorflowpip install keras # Possibly older release: conda install keras

Quick Start

Download Darknet model cfg and weights from the official YOLO website.
Convert the Darknet YOLO_v2 model to a Keras model.
Test the converted model on the small test set in images/.

wget http://pjreddie.com/media/files/yolo.weightswget https://raw.githubusercontent.com/pjreddie/darknet/master/cfg/yolo.cfg./yad2k.py yolo.cfg yolo.weights model_data/yolo.h5./test_yolo.py model_data/yolo.h5  # output in images/out/

See ./yad2k.py --help and ./test_yolo.py --help for more options.

More Details

The YAD2K converter currently only supports YOLO_v2 style models, this include the following configurations:darknet19_448, tiny-yolo-voc, yolo-voc, andyolo.

yad2k.py -p will produce a plot of the generated Keras model. For example seeyolo.png.

YAD2K assumes the Keras backend is Tensorflow. In particular for YOLO_v2 models with a passthrough layer, YAD2K usestf.space_to_depth to implement the passthrough layer. The evaluation script also directly uses Tensorflow tensors and usestf.non_max_suppression for the final output.

voc_conversion_scripts contains two scripts for converting the Pascal VOC image dataset with XML annotations to either HDF5 or TFRecords format for easier training with Keras or Tensorflow.

yad2k/models contains reference implementations of Darknet-19 and YOLO_v2.

train_overfit is a sample training script that overfits a YOLO_v2 model to a single image from the Pascal VOC dataset.

Known Issues and TODOs

Expand sample training script to train YOLO_v2 reference model on full dataset.
Support for additional Darknet layer types.
Tuck away the Tensorflow dependencies with Keras wrappers where possible.
YOLO_v2 model does not support fully convolutional mode. Current implementation assumes 1:1 aspect ratio images.

Darknets of Yore

YAD2K stands on the shoulders of giants.

0 0