How AI helped reconstruct a dead woman’s decomposed face, and a killer’s, to crack Nagpur rape-murder case | Mumbai News

How the rape-murder case was solved

The woman’s body was found on June 28 at a farmland in Patansawangi village under Saoner police station limits.

The police had initially registered a case of accidental death. However, a post-mortem revealed that the woman had been sexually assaulted and murdered. An FIR under Sections 64 and 103(1) of the Bharatiya Nyaya Sanhita (BNS) against an unidentified person was later filed by the police.

According to the police, identifying the victim was the first major hurdle. The investigation was carried out under the guidance of Nagpur Rural Superintendent of Police Harsh A. Poddar and Additional Superintendent of Police Anil Mhaske. The probe was led by Assistant Superintendent of Police and Savner Sub-Divisional Police Officer Sagar Kharde.

The challenges

Addressing a press conference, the police said the investigation was particularly challenging because the victim’s identity was unknown and there were no eyewitnesses, direct evidence or immediate clues about the perpetrator.

To investigate the crime, special teams comprising officers from the local crime branch, Saoner Police Station, Kelwad Police Station, and the Cyber Police Station were formed.

One of the breakthroughs came from the use of MARVEL’s AI-powered photo enhancement technology.

As the victim’s body had decomposed considerably, investigators used AI-assisted facial reconstruction to generate her probable appearance. To establish her identity, the reconstructed image was circulated through social workers and local networks.

During the investigation, ICJS and NATGRID platforms were used to analyse whether details of any missing person matched the victim’s description. Missing persons’ data from Nagpur city and Amravati police ranges, and neighbouring districts of Pandhurna, Chhindwara, and Seoni in Madhya Pradesh, were also scanned.

An important lead emerged after a social worker provided a photograph showing the woman alive on the Khaparkheda-Dehegaon road shortly before the crime.

The investigators had also recovered a chips packet wrapper from the crime scene, which led them to identify the shops selling the snack of that specific brand and analyse their online transactions.

Alongside conventional investigation, nearly 5,000 telecom data trajectories and mapping records from the area were analysed, and as many as 80 to 90 labourers were questioned in connection with the rape-murder case. The investigators also obtained and analysed registration details of nearly 15,000 vehicles from the Regional Transport Office to identify possible suspects.

How AI helped cops solve the case

Footage from 182 CCTV cameras installed across the area surrounding the crime scene was examined. Technical analysis of the CCTV footage between Pipla and Dehegaon helped the investigators identify the suspect. However, the available footage showed only a blurred image.

Using MARVEL’s AI Photo Enhancement technology, police reconstructed the suspect’s face and body profile.

The suspect was later detained in Nagpur city. When his actual facial features and body structure were compared with the AI-generated reconstruction, police said the system showed about 85 per cent similarity.

Sudhakar Dadarav Mandpe (48), a resident of Bansali Takli village in Saoner taluka, was arrested on July 6. He has a history of crimes against women. A court later remanded him to five days of police custody until July 11.