Improved Sensor-Based Animal Behavior Classification Performance through Conditional Generative Adversarial Network
This work addresses misalignment issues in dense prediction for animal behavior monitoring, offering incremental improvements for researchers in agricultural or behavioral sciences.
The paper tackled the problem of noisy and fragmented dense predictions in sensor-based animal behavior classification by modifying a U-Net with a Conditional Generative Adversarial Network (cGAN) and customized loss functions, resulting in improved accuracy from 92.17% to 94.66% on the UCI HAPT dataset and from 90.85% to 93.18% on pig data.
Many activity classifications segments data into fixed window size for feature extraction and classification. However, animal behaviors have various durations that do not match the predetermined window size. The dense labeling and dense prediction methods address this limitation by predicting labels for every point. Thus, by tracing the starting and ending points, we could know the time location and duration of all occurring activities. Still, the dense prediction could be noisy with misalignments problems. We modified the U-Net and Conditional Generative Adversarial Network (cGAN) with customized loss functions as a training strategy to reduce fragmentation and other misalignments. In cGAN, the discriminator and generator trained against each other like an adversarial competition. The generator produces dense predictions. The discriminator works as a high-level consistency check, in our case, pushing the generator to predict activities with reasonable duration. The model trained with cGAN shows better or comparable performance in the cow, pig, and UCI HAPT dataset. The cGAN-trained modified U-Net improved from 92.17% to 94.66% for the UCI HAPT dataset and from 90.85% to 93.18% for pig data compared to previous dense prediction work.