miguelmore / personality

Personality through Big Five from Selfies

Portrait Personality dataset

In folder "dataset" you can find the Portrait Personality dataset of selfies based on the ChaLearn dataset First Impressions. This dataset consists of 30,935 selfies labeled with apparent personality. Each selfie file was named with the prefix of the original video following by the frame's number, so in the file "bigfive_labels.csv" are stored all the labels where you can find the values of each trait of the Big Five model using the prefix (name of the original video).

Proposed Models

The main models are located in folder "models", per each model there are some files:

• Python notebook with the complete definition and process of the model.
• Preprocessed data (pickle files) for training and test into the folder "data_for_training&test".
• Best model got (model definition and trained weights) in folder "trained_model".

personality's Issues

An error occurred while loading the pickle file: [Errno 21] Is a directory: '/content/sample_data'

Train Model Files Corrupt

hi mate, hope you are doing well.
I just review your code and the dataset. It is fantastic work. But when I opened the train model(.rar) files that contains the weights of the model. it shows the files are corrupt. only Acoder(model) files works. I want to do some architectural work with the models. Please if you provide me the model(model of output of dense=5) file, it would be very helpful for me. And also saves my time by using your exceptional work.