Use tensorflow training SSD (a): configuration-related environment

When using TensorFlow target detection, you first need to download the tensorflow object detection API model Download this model is https://github.com/tensorflow/models. Followed by the need to complete the preparatory work, including: 1) to obtain research object_detection under protobuf due Object Detection API to use protobufs to configure and model training parameters, these files with the extension ".proto" placed models \ \ \ protos. Before using frames, you must use protobuf libraries compiled into py files in order to work properly. https://github.com/google/protobuf/releases select the corresponding version is installed protobuf Download. 2) Compile proto configuration file into the command line to models / research directory, execute the following command:. Protoc.exe object_detection / protos / * proto - python_out = If no information is displayed, it indicates that the operation was successful, this. when came down models / research / object_detection \ protos file find generated a lot py file. 3) detecting whether the API after normal completion of the operation, it is necessary to models \ research \ nets slim file folders copied to the models \ research; models \ research \ copy object_detection \ model_builder_test.py file to the builders models \ research under . Then run research folder model_build_test.py file, if the word OK appeared to show that everything is normal API, can be used normally.