Researchers develop AI method to detect sniffer dog success from tail movements

JERUSALEM, Aug. 26 (Xinhua) -- A multinational research team has created an artificial intelligence (AI) model that can determine when a sniffer dog detects a target scent with greater accuracy than professional dog handlers, Israel's University of Haifa said in a statement on Tuesday.

This new technology could provide handlers with real-time guidance, improving the effectiveness of sniffer dogs in security operations and life-saving missions, researchers said.

Detection dogs are widely used by police, customs, the military, and rescue teams because of their extraordinary sense of smell. They can detect substances that humans cannot, such as explosives, drugs, and cash.

Experienced handlers have long claimed they can anticipate a dog's success by observing subtle behaviors, particularly tail movements, but the argument had never been scientifically tested.

In the new study, published in Royal Society Open Science, researchers trained eight dogs to recognize a synthetic scent and indicate its presence by remaining still.

First, they searched for the scent at the concentration used during training. Second, the scent concentration was gradually reduced to make detection more difficult. All experiments were recorded with cameras that captured the dogs' movements.

The researchers used advanced computer vision technology to analyze tail angle, speed, and movement patterns. Based on this information, they developed an AI model to predict when the dog was near the target scent.

The new model achieved 77 percent accuracy, while 190 professional handlers identified only 46 percent of the cases correctly.

The researchers said that dogs show subtle signals before clearly indicating a scent, noting that the AI model detects these signals more accurately than the human eye. Enditem