Dlib "Error deserializing object of type short"

Question

I am getting error on pose_predictor = dlib.shape_predictor(predictor_model) on dlib python.

RuntimeError: Error deserializing object of type short
   while deserializing a floating point number.
   while deserializing a dlib::matrix
   while deserializing object of type std::vector
   while deserializing object of type std::vector
   while deserializing object of type std::vector

Tuque

this is very common error. you should download shape predictor model file and unpack it from archive. its also possible that file is corrupted while downloading, its size should be (unpacked) 99693937 bytes, sha1 = c7b0e0780a4fda180f75aadf654838c1866ee5fb – Brocky

Thanks @Brocky i'll try to download model again , and will verify with given sha1. – Tuque

Yeah @Brocky it worked <3 can u write this comment as answer ? so that i can accept the answer. – Tuque

Answer 1

I met the same issue and tried re-downloading the file, then it's OK now.

I downloaded from https://github.com/JeffTrain/selfie/blob/master/shape_predictor_68_face_landmarks.dat

Trapp

Answer 2

Download using command

wget -nd https://github.com/JeffTrain/selfie/raw/master/shape_predictor_68_face_landmarks.dat

when downloading from github download raw file instead of blob.

Neoteny

Answer 3

Hopefully this will help someone. I was using google colab and redownloading and reuploading did not work for me. So I cloned a git repo which had the shape predictor and used it's path

This is what I did

!git clone https://github.com/nicolasmetallo/eameo-faceswap-generator

cd eameo-faceswap-generator

Then change your shape predictor's path to

predictor = dlib.shape_predictor("shape_predictor_68_face_landmarks.dat")

Then run your code

Casie