Odakyu starts Panasonic fall-detection AI on station cameras

AI system

System: 転落検知システム (fall detection system): Panasonic image-analysis and deep-learning on existing station cameras
Model: Not assessed
Release: Not assessed
Application developer: Panasonic System Solutions Japan, with Odakyu Electric Railway
Role: Not assessed
Configuration: Not assessed
Inputs and outputs: Not assessed
Autonomy: Not assessed
Workflow: Existing station cameras covering the platform edge and track at two stations. Panasonic deep-learning image analysis trained from December 2019 to set detection areas and tell people from trains. A dedicated terminal in the station office that shows the alert and video for staff to act on.
Human oversight: Not assessed
Relation to AI: No event involving the system is documented. The account comes from Odakyu's own launch release and Impress Watch's report of it. They describe what the deep-learning analysis is meant to detect and how staff are alerted. The 2024 Asilla trial at two other stations is a separate system whose results were not reported.
AI contribution: Not assessed

Outcome identifiers: Not assessed
Benefit: From 1 April 2020 Odakyu's system analyzes existing platform camera video to flag people who fall onto the tracks or linger at the platform edge and alerts station staff; no outcomes are reported.
Category: Increased Product Safety
Status: Not assessed
Beneficiary groups: People. Organizations.
Beneficiaries: Not assessed
Reach: Not assessed
Distribution and access: Not assessed
Duration: Not assessed
Expected?: Yes
Costs and harms: Not assessed

Participants: Developer: Panasonic System Solutions Japan, with Odakyu Electric Railway. Operator: Odakyu Electric Railway.
Jurisdictions: Benefit: Kyōdō and Soshigaya-Ōkura stations, Odakyu Line, Japan. Developer: Japan. Operator: Japan.
Ownership and control: Not assessed
Dependencies: Not assessed
Cross-border connections: Not assessed
Geopolitical implications: Not assessed

Conditions for replication: Station staff present to confirm the alert and order an emergency stop. Detection areas calibrated per camera; the sources give no accuracy or false-alarm figures.
Replication attempts: 1 follow-up entries. Outcome unresolved: 1.
Later findings: 10 October 2024. Odakyu later extended AI analysis of existing station cameras, including detection of falls from the platform, to two more stations (Kakio and Tamagawa-Gakuen-mae) in a trial from 10 October 2024 to 9 January 2025, but with a different vendor's system (Asilla), and no result for either system has been reported.
Availability: Not assessed
Institutional decision: Not assessed
Grounds: Not assessed
Action and result: Not assessed
Success criterion: Not assessed

Assessed benefits: People. Organizations.
Depth: Not assessed
Reasons: People: No fall, alert or prevented accident is reported by any source. Organizations: The operator describes the intended staff workflow; no count of alerts, false alarms or staff time saved is published.
Observed or intended: Not assessed
Rubric: Not assessed
Assessor and date: Not assessed
Alternative assessments: Not assessed
Fill: Independently reported

Kind: Not assessed
Related entries: Not assessed
Field: Railway safety / platform fall detection. Station platform monitoring.
Purpose: Built to detect people falling from the platform onto the tracks or lingering at the platform edge and alert station staff. No detection event is documented.
Dates: Date basis unassessed: 2020 (precision unassessed). Updated: 21 October 2026.
Stage: Not assessed
Version: SHA-256: c44bc917ed3f…

News report Independent author

Impress Watch (2020). 既設の駅構内カメラで転落者自動検知。小田急. 31 March 2020.
News report Independent author

Impress Watch (2024). 小田急、駅構内カメラ映像に行動認識AI　転落や車いす、不審者検知. 7 October 2024 (the Asilla trial echo).
Press release Deployer

Odakyu Electric Railway (2020). 小田急電鉄、既設の駅構内カメラを活用した「転落検知システム」を運用開始. Press release of 30 March 2020, republished by Nikkei.