List of Publicly Available Pain Recognition Databases

This is a list of pain recognition databases that are publicly available for research. It has been started as part of the paper "Automatic Recognition Methods Supporting Pain Assessment: A Survey" [IEEEXplore | ResearchGate] to provide an overview on related databases and how to obtain them. The idea is to update this page whenever a new database is available or a database website URLs changes. Any contribution is welcome!

UNBC-McMaster Shoulder Pain Expression Archive Database

• Website
• subjects: 25 adults shoulder pain patients
• stimuli: 200 range of motion tests with affected and unaffected limbs
• data modalities: video of face (low resolution, includes social interaction / talking)
• annotation: self-report (VAS, sensory & affective verbal scales), observer-assessed pain intensity (OPI), affected/unaffected limb, FACS coding

BioVid Heat Pain Database

• Website
• subjects: 90 healthy adults (age 20-65)
• stimli: 14k heat pain (4 intensities × 20 repetitions × 2 parts × 90 participants); emotion elicitation, posed expression
• data modalities: video of face, EDA, ECG, sEMG (trapezius muscle; corrugator and zygomaticus for part B)
• annotation: stimulus (calibrated per person)

BP4D-Spontaneous Database

• Website
• subjects: 41 healthy adults (age 18-29)
• stimuli: 41 cold pressor task; emotion elicitation
• data modalities: video of face (color & 3D)
• annotation: stimulus, FACS coding

BP4D+ Database

• Website
• subjects: 140 healthy adults (age 18-66)
• stimuli: 140 cold pressor task; emotion elicitation
• data modalities: video of face (color, 3D, thermal), heart rate, respiration rate, blood pressure, EDA
• annotation: stimulus, FACS coding

MIntPAIN Database

• Website
• subjects: 20 healthy adults (age 22-42)
• stimuli: 2k electrical pain (40 stimuli in 4 intensities × 2 trials × 20 participants)
• data modalities: video of face (color, depth, thermal)
• annotation: stimulus (calibrated per person), self-report (VAS)

COPE Database

• Paper; to get the data, ask Prof. Sheryl Brahnam
• subjects: 26 neonates (age 18-36 hours)
• stimuli: 60 heel lancing for blood collection; non-painful stimuli
• data modalities: 204 photographs of face
• annotation: category (pain, rest, cry, air puff, or friction)

YouTube Dataset

• Paper; to get the data, ask Prof. Denise Harrison
• subjects: 142 infants (age 0-12 months)
• stimuli: immunizations (injection)
• data modalities: 142 videos with audio
• annotation: FLACC observer pain assessment

Infant Cry Signals Database (IIIT-S ICSD)

• Paper 1, Paper 2; to get the data, ask Prof. Vinay Kumar Mittal (see mail address in papers)
• subjects: 33 infants (age 3-24 months)
• stimuli: immunizations (injection) and other pain causes; non-painful cry causes
• data modalities: 693 audio cry samples
• annotation: category annotated by doctors and parents (pain, discomfort, hunger/thirst, and three others)

Neonatal Pain Assessment Database 1 (NPAD-1)

• Website, Paper
• subjects: 6 newborns and premature neonates
• stimuli: painful procedures (e.g., heel lancing and immunization)
• data modalities: videos with audio, vital signs
• annotation: pain assessment by two trained nurses with Neonatal Infant Pain Scale (NIPS); Gestational Age (GA), clinical data (e.g., the medication type/dose), birth-weight, race/ethnicity, gender, and non-pharmacological interventions (e.g., oral sucrose and the pacifier use)