北部地區固定污染源運用科技工具與AI智能化查核計畫
| 中文摘要 | 本計畫透過陳情案件、重大空氣污染案件之稽查裁處紀錄以及主辦單位提供相關資料，分析大園工業區以及觀音工業區之潛在污染源進行熱區圖層套疊與繪製，並分別於大園工業區提供６處架設參考地點以及於觀音工業區提供3處架設參考地點，共計９處可架設位置供主辦單位參考，以有效利用本案租賃之兩套雲端影像監視設備。 經主辦單位選定架設點位後，於１１２年４月２５日完成２組雲端影像監視設備架設，以及於１１２年５月３日空氣污染影像智慧辨識與異常影像資訊擷取功能正式上線。並於１１２年５月３１日完成平台建置，使用者可透過手機、平板、電腦進入平臺觀看即時影像、歷次異常影像以及近一個月之歷史影像調閱。而異常影像之ＬＩＮＥ推播通報功能亦於１１２年６月９日上線開始推播。統計至１１２年１１月３０日止，共計辨識出３,７８８則異常冒煙事件，後續經人工排查，其辨識成功率達９４．０%，透過每２個月提交之判煙辨識成果分析報告，以利稽查人員進行相關處置措施。 於大園、觀音點位達成蒐證之預期目標後，依主辦單位指示，配合於１１２年８月３０日與１１２年９月１１日變更雲端影像監視設備架設位置至觀音（更換方位）以及汐止，並持續實施自動化異常判煙蒐證與推播工作。期間也依合約規範，完成二次專案報告，並檢討與擬訂精進事項，供主辦單位參考。 | ||
|---|---|---|---|
| 中文關鍵字 | 固定污染源、影像辨識、環境監控 | ||
基本資訊
| 專案計畫編號 | 經費年度 | 112 | 計畫經費 | 1935 千元 | |
|---|---|---|---|---|---|
| 專案開始日期 | 2023/01/03 | 專案結束日期 | 2023/11/30 | 專案主持人 | 王御安 |
| 主辦單位 | 北區環境管理中心 | 承辦人 | 羅子亭 | 執行單位 | 振興發科技有限公司 |
成果下載
| 類型 | 檔名 | 檔案大小 | 說明 |
|---|---|---|---|
| 期末報告 | 112AA011.pdf | 33MB | 成果報告 |
Using technological tools and AI inspection of stationary pollution in the northern region
| 英文摘要 | This project uses public nuisance petition cases, major air pollution cases, inspection and punishment records, and relevant data provided by the Ministry of Environment to analyze the potential pollution sources in Dayuan and Guanyin industrial park and generate heat maps. We provide 6 locations for installation in Dayuan industrial park and 3 locations in Guanyin industrial park. In total, we provide nine installation locations to optimize the use of the two sets of cloud image monitoring equipment leased by this project. After selecting the installation points, two sets of cloud image monitoring equipment were installed on April 25, 2023, and the air pollution image intelligent recognition and abnormal image information extraction function was officially launched on May 3, 2023. And the platform was completed on May 31, 2023, users can enter the platform through mobile phones, tablets, computers to watch real-time images, previous abnormal images and historical images of the last month. The Line push notification function for abnormal images was also launched on June 9, 2023. As of November 30, 2023, a total of 3,788 abnormal smoke events were identified, and after manual screening, the recognition success rate reached 94.0%. Through the smoke recognition result analysis report submitted every two months, the inspection personnel can carry out relevant inspections measures. After achieving the expected goal of collecting evidence at Dayuan and Guanyin industrial park, according to the instructions of Ministry of Environment, the installation location of the cloud image monitoring equipment was changed to Guanyin (change monitoring direction) and Xizhi on August 30, 2023 and September 11, 2023, and the automated abnormal smoke detection and push work was continued. During the period, the second project report was completed in accordance with the contract specifications, and the review and improvement suggestions were also provided. | ||
|---|---|---|---|
| 英文關鍵字 | Stationary sources, Image recognition, Environmental monitoring | ||