Predatory behavior is crucial for survival and fitness in many species, evolving through complex interactions of genetics, adaptation, and learning. This interplay of innate and acquired neural mechanisms makes predation a rich subject for neuroscientific and neuroethological investigation. Current approaches to studying prey-capture use simple, barren setups that limit our understanding of how realistic environmental complexity impacts this process. This work presents both an experimental approach and analytical computational tools suitable to address these limitations. This study explores how predatory behaviors in mice adapt when introduced to a complex environment that simulates the human-adjacent environments in which mice and their prey thrive, such as landfills. Here, mice previously trained to hunt cockroaches in an empty bucket adapted to hunting in an enriched environment filled with refuse-like objects. We curated a new video behavior dataset and benchmarked/adapted various models for tracking and behavioral classification on new, specialized tasks. This approach allowed for the identification of novel predatory strategies such as skillful maneuvering through clutter, persistent prey tracking, and revisitation of prey hiding spots. These novel strategies were absent in the simpler experimental conditions widely used by this field. This work highlights the importance of studying animal behaviors in enriched environments to better understand how environmental complexity gives rise to adaptive behaviors.