SketchCode是一种深度学习模型，它采用手绘的Web原型并将其转换为有效的HTML代码，使用图像字幕体系结构从手绘网站线框生成HTML标记。
注意：此项目是概念验证；该模型性能依赖于类似于核心数据集的线框。
依赖
Python3(notcompatiblewithpython2)pip安装依赖
pipinstall-rrequirements.txt用法示例下载数据和预训练的权重：#Gettingthedata,1,700images,342mbgitclonehttps://github.com/ashnkumar/sketch-code.gitcdsketch-codecdscripts#Getthedataandpretrainedweightsshget_data.shshget_pretrained_model.sh使用预训练的权重将示例绘制的图像转换为HTML代码：cdsrcpythonconvert_single_image.py--png_path../examples/drawn_example1.png\--output_folder./generated_html\--model_json_file../bin/model_json.json\--model_weights_file../bin/weights.h5一般用法使用权重将单个图像转换为HTML代码：cdsrcpythonconvert_single_image.py--png_path{path/to/img.png}\--output_folder{folder/to/output/html}\--model_json_file{path/to/model/json_file.json}\--model_weights_file{path/to/model/weights.h5}将文件夹中的一批图像转换为HTML：cdsrcpythonconvert_batch_of_images.py--pngs_path{path/to/folder/with/pngs}\--output_folder{folder/to/output/html}\--model_json_file{path/to/model/json_file.json}\--model_weights_file{path/to/model/weights.h5}训练模型：cdsrc#trainingfromscratch#<augment_training_data>addsKerasImageDataGeneratoraugmentationfortrainingimagespythontrain.py--data_input_path{path/to/folder/with/pngs/guis}\--validation_split0.2\--epochs10\--model_output_path{path/to/output/model}--augment_training_data1#trainingstartingwithpretrainedmodelpythontrain.py--data_input_path{path/to/folder/with/pngs/guis}\--validation_split0.2\--epochs10\--model_output_path{path/to/output/model}\--model_json_file../bin/model_json.json\--model_weights_file../bin/pretrained_weights.h5\--augment_training_data1使用BLEU分数评估生成的预测cdsrc#evaluatesingleGUIpredictionpythonevaluate_single_gui.py--original_gui_filepath{path/to/original/gui/file}\--predicted_gui_filepath{path/to/predicted/gui/file}#trainingstartingwithpretrainedmodelpythonevaluate_batch_guis.py--original_guis_filepath{path/to/folder/with/original/guis}\--predicted_guis_filepath{path/to/folder/with/predicted/guis}
评论