search

ivineetm007 / drunk-detection

This repository contains the source code for the paper Dataset of Perceived Intoxicated Faces for Drunk Person Identification.
MIT License
11 stars 5 forks source link

VIdeo model help #3

Open Kushagra0409 opened 11 months ago

Kushagra0409 commented 11 months ago

I am currently using the VGG-LSTM model for drunk video detection, and I utilized the extracted features to get a model. However, when using my own video at 10seconds, 244x244 24fps .mp4 format, and doing features extraction on that video (using vgg-face), I am getting misclassification problems. Could you please Guide if I missed any steps (there was mention of video preprocessing pipeline github, but I did not go through that, I directly performed vgg-face7 feature extraction on my raw videos, as given in the notebook).

Jay-sanjay commented 7 months ago

Hi @Kushagra0409 and @ivineetm007 my self Jay BTech Comp Engineering, IIT-Hyderabad. Can you please help me getting with the dataset, that was used to train and test this project, I am really interested in it .