/EmteqActivityRecognition

Emteq Human Activity Recognition Challenge 2019

Primary LanguageJupyter Notebook

Emteq organized Human Activty Recognition Challenge 2019. The main theme of this challenge is to recognize different types of activities from sensor data using any machine learning model.

Dataset

  • There are 4 persons' data. 3 persons' data is used as train data and 4th person's data is used as test data
  • Dataset download Link : Emteq Dataset

Tools

  • Jupyter Notebook

Description

  • 1D CNN is used as the deep learning model

  • Actity chart

  • Confusion Matrix

  • Classificatin Report

1D CNN Model

model = Sequential()
model.add(Conv1D(filters=64, kernel_size=3, activation='relu', input_shape=(n_timesteps,n_features)))
model.add(Conv1D(filters=64, kernel_size=3, activation='relu', padding = 'same'))
model.add(Dropout(0.5))
model.add(MaxPooling1D(pool_size=2))
model.add(Flatten())
model.add(Dense(100, activation='relu'))
model.add(Dense(n_outputs, activation='softmax'))

Prize

  • 2nd prize was won from Sozo Lab team. Result