Shoplifting Dataset (2022) - CV Laboratory MNNIT Allahabad

Synthesized shoplifting dataset is created by ourselves in the CV laboratory of MNNIT Allahabad with the help of 32 Megapixel camera. The recorded videos are of 640x480 resolution and captured at frame rate of 30 frames per second. This dataset has clearly distinguishable person's normal and stealing actions. Our self-created dataset consists of two classes:
1) Normal
2) Shoplifting
The normal class contains video clips that manifest activities like walking, inspecting shop items, etc. Other side, the shoplifting class has video clips representing object stealing actions by human being like putting store items under clothing or bags etc.

Note: Dataset required pre-processing step to clean video scenes