Tensorflow Based Interview Analysis

 Tensorflow Based Interview Analysis :

Introduction:

• Tensorflow Based Interview analysis is a project which is built to provide an end-to-end AI interviewing system that will be develop using asynchronous video interview processing and a TensorFlow AI engine to perform automatic personality recognition based on the features extracted from the asynchronous video interview and the true personality scores from the facial expressions. 

• It will analyze the Uploaded resume and recommend the perfect role for the candidate appearing for the interview.

• The tone of the candidate will also be taken into consideration and by analyzing the tone, feelings of the candidate will be shown.

• The facial expression of the candidate will also be taken care of.

• Additionally, there is signup and Login for the Interviewee and also Registration for the Interviewer.

 Objectives:

1. Video-based face feature analysis: With the help of the CNN algorithm and face landmark user face will be captured by the camera and feature extraction of the face will be done which will result in obtaining output such as the happy face, entertaining face, good gesture, good smile. Then effectiveness with the job location will be identified.

2. CNN model generation: face categories will be added and trained with convolution neural network

3. Text model generation: Random forest, naïve Bayes, SVM model will be prepared for tone analysis.

4. Face recognition:-for employee verification we can use a face recognition module associated with azure Microsoft

Flowcharts :

• Registration of company employee:

•  Resume Verification :

• Tone/Text analysis :

• Face analysis :

• Flowchart

Technology Stack :

Website :

• HTML

• Javascript

• Bootstrap

• python

• Microsoft Visual Studio Code

• Google chrome

Backend:

• Mysql

• Python

• OpenCV

• Matplotlib

• Kaggle 

• NumPy

• Tensorflow

• Atom 

 

    

Data Flow :

Conclusion:

• This project is for personality computing. In traditional personality computing, validating APR using manually labeled features from any possible detectable distal cues was quite complicated.

• This project developed an AVI embedded with a TensorFlow-based semi-supervised DL model to accurately auto-recognize an interviewee’s true job applicants. Our APR approach achieved an accuracy above 90%, outperforming previous related laboratory studies whose accuracy ranged between 61% and 75% in the context of nonverbal communication. The high-performing APR used in this AVI can be adopted to supplement or replace self-reported personality assessment methods that can be distorted by job applicants due to the effects of social desire to be selected for employment. 

• Previous related studies have found that multimodal features (image frames and audio) learned by deep neural networks can deliver better performances in predicting the big five traits than can unimodal features

Acknowledgment:

 It gives me immense pleasure to submit the mega-project blog on Tensorflow Based Interview Analysis to my guide, Professor Chandrajeet Borkar, and Head of Department Dr. Latesh Bhagat, Computer Science and Engineering department who was a constant source of guidance and inspiration through the seminar work.

 

    I am also very thankful to all the Computer Science and Engineering departments; whose encouragement and suggestions helped me complete the Mega-Project work.

 

    I also express my sincere gratitude to our respected principal, Dr. Nitin Ghawghawe, to provide us with the necessary facilities.

 

    At last, I am thankful to my friends whose encouragement and constant inspiration helped me to complete this seminar work verbally and theoretically.

 

 

 

 

 

Project Details :

• College: Government College of Engineering Nagpur

• Department: Computer Science and Engineering

• Department HOD: Dr. Latesh Bhagat

• Project Guide : Prof. Chandrajeet Borkar

• Group Members :

• Ankit Agarkar 

• Purva Deshpande