![]() ![]() It is a random division of the training set and the verification set of the picture. Where the code segment: train_anno, val_anno = train_test_split(anno, test_size=0.1) Val_anno.to_csv('CSV/val_annotations.csv', index=None, header=None) Train_anno.to_csv('CSV/train_annotations.csv', index=None, header=None) Train_anno, val_anno = train_test_split(anno, test_size=0.1) # anno.to_csv('CSV/annotations.csv', index=None, header=None) Img_path_list.append(restrict_rele_path+img_name)Ĭategory_list.append(anno_collect-1)Īnno = anno.map(mapper) data/imgs/img_002.jpg,215,312,279,391,catīelow, I posted a code I wrote: def restrict_image_info(label_path): If an image does not contain any objects to detect, the format is as follows: path/to/image.jpg,Ī complete example: /data/imgs/img_001.jpg,837,346,981,456,cow (2) According to the example of the official website, the format of the Annotations data set produced by itself is as follows: path/to/image.jpg,x1,y1,x2,y2,class_name |_ data # (optional), so annotations.csv can use the relative path of the image ![]() (1) at keras-retinanet-master/keras_retinanet/Create a new folder below the folder CSVUsed to store your own data sets. (Of course, it is best to have the corresponding version of the full C++) If the error prompts directly to Baidu, there will be a solution. There is no compiler for a version of C++. (3) Model compilation can use the following commands: python setup.py build_ext -inplace Here to mention, if a package download installation is not successful at the time of installation, you can write down the version, such as opencv-python 3.4.5.20, you can directly use pip or conda to install, but you must remember the corresponding version. (1) Model download address: fizyr/keras-retinanet (2) Model installation can use the following command: pip install numpy -userĭuring the installation process, the dependencies are checked, such as opencv-python, which is loaded and installed if it is not installed. 1, the code open source framework is using fizyr/keras-retinanet 2, Keras version should be 2.2.4 or higher ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |