Datasets:
File size: 4,071 Bytes
6c1c833 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 | {
"field_id": "F027",
"field": "biometric recognition",
"directions": [
{
"direction_id": "G0",
"name": "Facial identity learning and robust matching",
"scope": "Learn or obtain evidence for distinguishing facial identities, and maintain identity matching when real training data is limited or facial appearance or image quality changes. Includes the construction and use of synthetic identity data."
},
{
"direction_id": "G1",
"name": "Demographic equity in face recognition",
"scope": "Measure, explain, or reduce differences in face-recognition outcomes across demographic groups."
},
{
"direction_id": "G2",
"name": "Person re-identification across observations",
"scope": "Find the same person across different image or video observations by learning and retrieving identity cues, while addressing matching difficulties arising from target domains, unseen domains, visible–infrared differences, partial occlusion, clothing changes, and temporal observations."
},
{
"direction_id": "G3",
"name": "Identity from gait and performed behavior",
"scope": "Identify or verify people from walking structure and dynamics, sequences of device interactions, or handwritten signatures. Gait recognition includes the full range of silhouette, skeleton, and body-shape cues."
},
{
"direction_id": "G4",
"name": "Ocular and fingerprint identity matching",
"scope": "Extract and compare identity information from the eye region or fingerprints for verification or identification, including variation in acquisition, localization, and latent fingerprint matching."
},
{
"direction_id": "G5",
"name": "Fusion of complementary biometric identity cues",
"scope": "Improve person recognition by combining distinct identity traits, sensors, or semantic cues, including body, audio, visual, finger, and soft-biometric evidence. Fusion contributes to the identity decision rather than serving only to enhance an image."
},
{
"direction_id": "G6",
"name": "Restoration of biometric imagery for recognition",
"scope": "Recover image information useful for identity matching from degraded face, fingerprint, iris, or vascular images."
},
{
"direction_id": "G7",
"name": "Biometric sample quality and recognition utility",
"scope": "Assess biometric sample quality and the effects of acquisition or encoding degradation on identity-matching utility, without requiring a new matcher."
},
{
"direction_id": "G8",
"name": "Detection and mitigation of biometric authentication attacks",
"scope": "Detect spoofed biometric presentations or face images blending multiple identities, and strengthen identity authentication against adversarial inputs, active behavioral attacks, and training-data poisoning."
},
{
"direction_id": "G9",
"name": "Biometric evasion and impersonation attack construction",
"scope": "Construct or test inputs that cause an identity matcher to miss an identity, misidentify someone, or accept an impostor. Includes digital and physical perturbations and face images intended to match multiple contributing identities."
},
{
"direction_id": "G10",
"name": "Biometric template privacy and information leakage",
"scope": "Limit disclosure and cross-use of biometric images, features, or templates while preserving their use for identity matching, and test whether representations or matcher outputs still reveal private information."
},
{
"direction_id": "G11",
"name": "Explanations of biometric matching decisions",
"scope": "Explain which image regions and learned features support a specific identity match or non-match decision."
},
{
"direction_id": "G12",
"name": "Resource-efficient biometric recognition",
"scope": "Reduce model, inference, or retrieval costs while retaining useful identity-matching capability."
}
]
}
|