Daily Routine Activity Video Dataset

I’m assembling a fresh video dataset focused on activity recognition and I need original footage of real-world daily routines. The goal is to train computer-vision models, so every clip must clearly show the action and end with enough “idle” frames for clean labeling. What I’m looking for • General actions such as walking through a hallway, sitting down, eating, opening doors, using a phone. • Specific tasks like brushing teeth, cooking a meal, tying shoelaces, folding laundry, making coffee, watering plants. Feel free to capture these moments indoors or outdoors, in different lighting conditions, camera angles, and with varied backgrounds. Diversity of age, gender, and setting is strongly encouraged. Recording guidelines • Shoot in 1080p or higher, landscape orientation. • Keep each clip between 10–60 seconds. • Only one clearly visible activity per clip. • No copyrighted TV/music in the background and no sensitive personal info on screen. • Save each file as MP4 or MOV and name it with the activity (e.g., “brushing_teeth_03.mov”). • Provide a simple CSV with filename, activity label, location type (kitchen, park, etc.), and any notable conditions (low light, crowded). Deliverables 1. A minimum of 30 labeled clips that meet the guidelines. 2. The accompanying CSV metadata file. Acceptance I’ll review a small sample first; once quality is confirmed, you can proceed with the rest. Clips that are blurry, overly shaky, or include restricted content will be rejected. Future work Although this round is limited to daily routines, I may later expand to sports events or workplace tasks, so strong contributors could see repeat projects. If you have the gear and can capture authentic everyday moments, let’s get started—looking forward to seeing your footage!