-
HW1: In this homework, you are required to train and test a Multimedia Event Detection (MED). More specifically, your MED will be able to detect three different events: assembling a shelter, batting in a run and making a cake. You will build two different MED. The first one will use Mel-Frequency Cepstrum Coefficients (MFCC) and the second one will use Automatic Speech Recognizer (ASR) transcriptions.
-
HW2: In this homework, you are required to train and test a Video-based Multimedia Event Detection (MED). As before, your MED will be able to detect the three events: assembling a shelter, batting in a run and making a cake. As the previous homework, you will build three different binary classifiers. The first one will use Improved Trajectory Features (Imtraj), the second one Scale-Invariant Feature Transform (SIFT) and the last one will use features extracted from a Convolution Neural Network (CNN).
-
HW3: This is the final homework of the course. In this homework, you are required to build an complete Multimedia Event Detection (MED) using the material generated in homework 1 and homework 2. You need to present three different classification solutions and report their results. You are free to use any of the fusion techniques learned in lectures.