{ "name": "Gilt_posture_dataset", "version": "1.0", "description": "A multimodal dataset for pig posture detection using synchronized RGB and depth images. Includes labels for five pig postures, real-world object height from the floor, and the metadata.", "contributors": [ { "name": "Anil Bhujel", "email": "bhujelan@msu.edu", "affiliation": "Michigan State University", "role": "Research Associate" } ], "date_created": "2025-05-25", "license": "CC BY 4.0", "citation": { "title": "A Computer Vision dataset for Gilts' daily activity monitoring and tracking", "authors": ["Anil Bhujel", "Daniel Morris", "Janice Siegford", "Madonna Benjamin", "Mk Bashar"], "year": 2025, "doi": "To be assigned", "url": "yet to assigned" }, "modalities": [ { "type": "color", "description": "RGB images from a ceiling-mounted RGB-D camera.", "height": "2.4384 meter", "camera angle": "approx 40 deg", "resolution": "1920x1080", "frame rate": "30", "format": "PNG" }, { "type": "depth", "description": "Depth images aligned with color, in millimeters.", "resolution": "640x576", "frame rate": "30", "format": "PNG" }, { "type": "height_map", "description": "Object height from the floor estimated using a floor plane model." } ], "annotations": { "posture_labels": ["Feeding", "Lateral_lying", "Sitting", "Standing", "Sternal_lying"], "format": "Yolov11 .txt", "labeling": "Bounding box" }, "data_split": { "train": 0.75, "val": 0.15, "test": 0.1, "criteria": "random" }, "collection": { "sensor_setup": "Orbbec Femto Mega RGB-D cameras mounted at ~8Ft height, angled ~40°", "camera_parameters": "provided as orbbec_parameters.yaml file", "frame_rate": "30 FPS", "synchronization": "Synced with Orbbec sync hub pro", "depth_to_height_conversion": "Using calibrated floor plane model", "rfid_triggered": true, "date_of_recording": "01/08/2025", "recording_window": "07:00 to 14:00", "recording_duration": "Continuous" }, "environment": { "pen_type": "MSU swine research center", "pen_size_meter": [4.8768, 2.4384], "pen_fence_height_meter": 0.825 "floor_type": "Slatted wood", "lighting_conditions": "Artificial, stable intensity", "ventilation": "Mechanical ventilation with air flow" }, "animals": { "breed": ["Purebred Yorkshire", "Exotic-crossbred", "Meishan"], "number_of_animals": "18 (9 per pen)", "age_range_weeks": 24, "expected_delivery_date": "03/08/2025", "tagging_method": "RFID and ear tags" }, "download": { "url": "https://huggingface.co/datasets/anilbhujel/Gilt_posture_dataset", "file_format": "ZIP", "size": "6.7 GB" } }