Economic migration is a critical issue for many European countries and disasters involving migrants present a related global challenge [1]. Migrant disasters can be defined as the multiple deaths of migrants caused by catastrophic circumstances whilst crossing bodies of water or land between origin and destination. These include capsized boats, wildfires, overcrowding (leading to suffocation) and extreme weather conditions. A disaster may include mass casualties in a single incident (e.g., a capsized boat) or multiple deaths over time with the same pattern (e.g., exposure on migrant land routes). Therefore, where migrants travel great distances across and between continents, a large percentage reach the end of their lives, and, whilst sometimes their bodies are found, unfortunately many are lost at sea, in deserts or in wasteland. As a result, many families may never learn of the fate of their relatives, and few migrants who perish are ever positively identified.

In recent years, due to war and socioeconomic crises, there has been an increased use of dangerous maritime routes to Europe across the Mediterranean and via the Canary Islands and hostile land routes to Europe through Turkey, and a related increase in the number of unidentified deceased migrants [2]. However, the number of fatalities that involve migrants fleeing from countries with volatile political and socioeconomic situations is difficult to calculate [3]. The International Organization for Migration (IOM) has estimated that more than 60,000 migrants have died on these journeys with the majority of disasters taking place in the Southern European borders [4], where Italy, Spain, Malta and Greece are the most affected countries [5,6,7]. On a smaller scale, there has been a significant increase in migrant deaths along the borders between Mexico and the United States in the last decade [4].

In order to address these challenges, the European Cooperation in Science and Technology (COST) Action CA22106 [8] focusing on Migrant Disaster Victim Identification (MDVI), signifies a concerted European effort to enhance inter-agency communication along with new developments in scientific and technical capabilities in forensic science and humanitarian work. COST aims to develop new, efficient methods for identifying deceased migrants by facilitating cooperation and knowledge exchange among researchers, NGOs, forensic experts, and other stakeholders across Europe, thereby addressing this pressing humanitarian issue more effectively. This initiative not only seeks to improve identification processes, but also by ensuring that families can learn the fate of their loved ones, support human dignity and uphold the human rights which are at the forefront of European values.

The IOM states that migrants’ rights should be protected during the whole circle of migration [9], and this is especially challenging where the individuals originate from low-income countries, where traditional so-called primary identifiers are limited or absent, and migration is undocumented or unmonitored. The identification of migrants has been defined by identification experts [10] as an “enormous paradox”. It is recognised by the Geneva Convention that identifying a deceased person is a fundamental right and a moral obligation that should be respected without discrimination [11]. However, while identifying the deceased in a mass disaster context is standardised, regulated by internationally accepted procedures and conducted by adequately trained personnel, the identification process becomes slow and lacks urgency in the aftermath of disasters involving migrants [12, 10]. This has led to an incredibly large number of unidentified victims who are buried without a name, leaving families to deal with the psychological consequences of not knowing the fate of their missing loved ones [3]. From a humanitarian and legal point of view, the lack of identification for these victims can be considered a violation of human rights that does not only affect the dignity of the deceased but largely affects the rights of the living [10, 13, 14] (Fig. 1).

Distribution of post-mortem-ante-mortem (PM-AM) cases with different AM image quality and related level of support for a match

Migrant mass disasters include some unique challenges that are different from many ‘usual’ DVI events, although there are similarities with other humanitarian identification challenges, such as the disappeared in countries such as Guatemala, Timor-Leste and Argentina. In these humanitarian identification cases, primary methods of identification are often inappropriate or even impossible, due to legal obstacles and socioeconomic conditions of the countries of origin, fear of reprisals or financial consequences for the migrants’ families and the resulting lack of regular dental treatment, fingerprint resources and medical records. In addition, water-damage, rapid decomposition and delayed access to the migrant bodies may further confound any comparative methods [15]. The IOM states that the majority of deceased migrants are identified visually by family members [16], despite international recognition of significant problems associated with visual recognition of the dead, due to psychological/social pressures and post-mortem (PM) changes [17]. Recently, the division between primary and secondary identifiers has been under discussion in international policymaking groups, such as INTERPOL DVI Working Group.Footnote 1 In a statement published by Forensic Anthropology Society of Europe [18], members argued that the preferred primary identifier cannot always be used to support identification and often secondary identifiers provide sufficient evidentiary value for a conclusive identification with the use of a reliable methodology [18]. There are also significant challenges in relation to the collection of ante-mortem (AM) data from families of the missing in the countries of origin. Since many migrants attempt crossings via trafficking gangs, communication with relatives of the missing may create danger for the family, and relatives may be reluctant to speak to the authorities due to fears of criminal investigation or social exclusion [7].

Craniofacial analysis offers valuable tools for forensic investigation [19, 20], but these techniques have not been widely utilised in MDVI, due to the INTERPOL DVI recommendation for the utilisation of primary identifiers. The contemporary rise of global networks and the use of mobile phone images has led to public acceptance of social media use in the search for missing people; platforms, such as Facebook and Instagram, are frequently utilised to publish AM images of missing people [21]. This has created an opportunity for digital facial analysis to be employed to match PM images to AM images. In the aftermath of a disaster, critical evidence that migrants carry with them, including mobile phones, is often neglected. Much of the useful identification data stored on mobile phones can be accessed in social media or cloud storage, but this data is hardly ever collected, as it would likely involve interviews with families in the country of origin to confirm social media ownership [16] and require permission to access image data. Some European forensic databases have utilised 3D imaging for craniofacial analysis, but these tend to be limited to national forensic casework rather than large-scale disaster scenarios [22, 23].

Facial identification is described as a meticulous, systematic and well-documented manual approach [24, 25] and generally involves comparison between the face of an individual visible in photographs or video frames and facial image/s of the suspected identity [26]. It has become particularly relevant in many courtrooms across the globe due to the increased use and availability of Close Circuit Television (CCTV) footage worldwide [27]. Morphological comparison involves comparing facial images, to assess the similarities and differences by examining the correspondence between shape, appearance, and location of features, as well as identifying features like scars and moles [25, 28, 29]. Facial image comparison experts working for law enforcement, government organizations, and academic institutions and are routinely utilised in legal scenarios. These experts are usually trained in anatomy, principles of image comparison and recognition, and court duties [30]. Their role involves providing expert opinion on the identity of an individual or their potential association with a criminal event by comparing facial images collected at the scene of crime (CCTV or security footage) with those of suspects [31].

Facial image comparison is based on the assumption that the combination of morphological traits observable on faces is unique [25]. However, subjectivity is inherent [32] as the face and its features and shapes are difficult to assess quantitatively [24] due to expression, ageing, facial modifications, image qualities and distractions. Therefore, the final decision is based on the description of similarities and differences and includes a scale of support to express the level of confidence in a match or exclusion. The international joint work of the Facial Identification Scientific Working Group (FISWG) has produced guidelines and protocols for facial comparison [33], including a comprehensive list of facial features to analyse, image quality factors to consider and recommendations for training and competency for experts [34, 35]. The FISWG morphological features list is currently considered the most comprehensive and relevant protocol for the forensic facial comparison [36]. Tests have been carried out to validate the accuracy of the morphological comparison using FISWG guidelines on living individuals; a real-life scenario was simulated with various image types and results suggest an accuracy of 71–99%, with image quality having the greatest impact on accuracy [31]. Currently, there is no minimum requirement for image quality and the images are usually assessed for suitability by the practitioner against factors such as pixel resolution, distortion, pose, lighting and obstruction caused by objects such as hats and sunglasses [37]. The relationship between lower match accuracy and low quality of the image has been confirmed [31, 38,39,40,41], and higher accuracy is produced by the combination of morphological analysis and optimal image quality [26, 31]. The accuracy of facial image comparison is considered dependent upon the experience, knowledge and ability of the practitioner [42].

In recent years, there has been a growing interest in using morphological comparison for PM identification [43] particularly for the increasing number of unidentified bodies associated with migrant disasters and other mass fatalities [44]. Images are often a crucial piece of personal data that is easily available to investigators; PM images and descriptions of external morphological features of both the body and face are well-documented by pathologists and first responders [10, 45]. Also, the availability of personal images has significantly increased with the widespread use of social media platforms and smartphones equipped with cameras, meaning that often there is easy access to high-quality images of the faces of missing people. In the context of migration, researchers [44] have highlighted the use of images provided by families to aid in identification and recent MDVI experts in Italy have reported that facial image comparison, particularly where moles and scars were visible, played a fundamental role in supporting DNA analysis for achieving identification [12]. Generally, these elements are listed as comparable facial features that should be analysed during identification of living individuals, but they appear to assume a more significant role in the identification of deceased individuals.

However, the application of facial identification methods to deceased individuals has not received the same consideration as for identification of the living, and this may be due to a lack of knowledge relating to facial decomposition patterns [46]. Decomposition is a natural process that affects an individual after death and gradually leads to the destruction of the tissues [47] following a 5-stage process: fresh, bloat, active decay, advanced decay, and skeletonization [48]. However, these stages are heavily influenced by many intrinsic and extrinsic factors. Intrinsic factors include weight and size of the body, posture of the body at death, and clothing. Extrinsic factors include the environment, temperature, weather, humidity, type of burial, level of insect, and other animal activity [49,50,51]. PM changes are generally divided into early, mid and late changes. Early PM changes usually appear within 2 h from death, and are characterised by biomolecular changes, visible skin pallor due to cessation in blood circulation, as well as the purging of stomach material and general flaccidity [47]. Mid PM changes include algor mortis, rigor mortis and livor mortis, and later PM changes include putrefaction, mummification and skeletonisation. Only two studies have attempted to quantify the degree of early decomposition changes that affect the face: one in the field [52] and one in a mortuary setting [46]. These studies used 3D laser scans to evaluate patterns of facial decomposition, and, despite the different environments and timelines, both determined the major changes as bloating at the lateral parts of the face, swelling at the orbits with darkening of the eyeballs, and shrinkage at the midline region of the face and upper lips.

The term “skin mark” encompasses a variety of skin irregularities such as moles (or nevi), scars, discolouration, acne, eczema lentigines, cherry hemangiomas, and seborrheic keratoses, with some of them being less stable through time [53]. Skin marks such as moles, blemishes and scars can be powerful morphological elements that provide an important support for facial identification [54, 55]. Assessing the nature of skin marks from a photograph, especially if photographs are of low quality, is difficult, and experts suggest using only recent photographs for skin mark comparison. Researchers [56] have examined the significance of position, shape, and incidence of scars and nevi on the dorsum of the hand and suggest that these characteristics can be crucial in identification. Scars and tattoos were accepted as the sole means of identification in 2% of the victims of the aeroplane disaster in Linate, Milan, 2001 [57] and the early stages of identification in the terroristic attacks in Paris between 2015 and 2016 were enhanced by using strong secondary identifiers, such as tattoos and scars [58]. A similar approach has been utilised in challenging migrant identifications in Greece, where the initial recognition of personal belongings through photographs provided by family members, served as indicative identification later confirmed by genetic kinship analysis [6]. Reviews of DVI procedures [59] in low-resource countries, emphasise the importance of strong secondary identifiers, particularly in situations where primary methods cannot be used. These facial marks are usually recorded during the PM analysis of the victims; however, their stability can be influenced by the decomposition process [17]. Facial marks, such as moles and scars and any blemish visible on the skin, are among the most common features across different populations and these are often visible in photographs [56, 60, 61]. In the same way, facial lines or creases have been shown to be consistent enough to be used for human identification [62] even with early decomposition effects, such as bloating.

Another element of the face that has been found particularly unique is the human ear. The stability of this feature throughout a lifetime, as well as the stability maintained with different facial expressions [