How environmental features (e.g. people or other animals) affect movement is an important element for the study of animal behavior, biomechanics and welfare. Here we present a stationary overhead camera-based persistent monitoring framework for the investigation of bottlenose dolphins (Tursiops truncatus) response to environmental stimulus. Mask R-CNN, a convolutional neural network architecture, was trained to detect 3 types of targets in the environment: dolphins, people, and enrichment floats intended to stimulate and engage the animals. Detected targets within each video frame were linked together to create track segments across frames. The animal tracks were used to parameterize animal response to the presence of environmental stimuli. We collected and analyzed over 10 hours of data, 4 hours with and 6 hours without enrichment, from animals in a manage lagoon environment. With no enrichment present, animals swam with an average speed of 1.08 m/s, and tended to be attracted to areas of the environment with people present. Floating enrichment in the environment created a dynamic point of attraction, lowering the average swimming speed to 0.64m/s and modifying the use of the full environment. These preliminary results indicate that the presence of enrichment and people in, or around, the environment attract the animals, biasing habitat use and movement patterns as a result. This work demonstrates the ability of the proposed framework for the quantification and persistent monitoring of bottlenose dolphins, and will enable new studies to investigate individual and group animal locomotion and behavior.