REAL TIME OBJECT DETECTION USING TENSORFLOW

I was going through the implementation on "ssd_mobilenet_v1_coco_2017", the part I was really getting stuck on was protobuf compilation for os. What worked for me though is first downloading the system relevant protoc file from

https://github.com/google/protobuf/releases

Exporting protoc through

export PATH=$PATH: (path to your protoc file) in the terminal window.

Then going into the models/research/object_detection/protos and compiling each and every file, in short

protoc *.proto --python_out=. was not working for few files, then instead of *.proto, I directly put all the file names required for the work and compiled it and it did work.

As Mac doesn't have a GPU, thanks to apple for costly pricing and no GPU, fps was really really slow, though it worked!!

I converted the notebook to .py, I always prefer to do it through the command window in windows or terminal in macos as if something is wrong, you can see it the terminal.

You can find the code here: Github Code