紅外線自動相機-野生動物觀測影像

目前在林業試驗所生態感測平台http://iesn.tfri.gov.tw/forestDW/針對蛙類聲景監測、珍稀植物菱形奴草線上直播監測、以及野生動物監測在蓮華池研究中心農委會林業試驗所的試驗林一號及二號量水堰之間，設置一處面積25公頃(500m×500m)之蓮華池闊葉林長期動態樣區，並於區域內架設紅外線自動相機進行野生動物的監測。目前經這幾年各個錄影機錄影拍攝到的動物資料哺乳類有：山羌、石虎、獼猴、穿山甲、白鼻心、赤腹松鼠、野豬、食蟹獴，鳥類有：藍腹鷴、竹雞、深山竹雞、黑冠麻鷺、虎鶇、翠翼鳩。資料收集方式以動物經過觸發錄影30秒後停止錄影，錄影間隔為每5分鐘不重複錄影，以防止因動物重複來回觸發錄影造成檔案過多，記憶卡容量不足與電池過度消耗而無電力。目前資料收取間隔以一個月人員上山更換記憶卡與充電電池收取回的影像檔以人工的方式進行自動相機編號歸檔，影像拍攝日期建立檔名。然後再經過人工的方式一個一個影像檔一一觀看，進行人工動物分類辨識作業再將分類好的資料以動物名稱以日經過次數、月經過次數、年總數量作為日後分析的數據。At the Ecological Sensing Platform of the Taiwan Forestry Research Institute (TFRI, Council of Agriculture, Executive Yuan) hosted at http://iesn.tfri.gov.tw/forestDW/, a 25-hectare (500 m × 500 m) permanent broad-leaved forest dynamic plot has been established in the experimental forest of the TFRI Lianhuachi Research Center, between Weir No.1 and Weir No.2. This plot is designed for three monitoring tasks: amphibian acoustic environment monitoring, online live-stream monitoring of the rare plant *Mitrastemon kawakamii*, and wildlife monitoring. Infrared automatic cameras have been deployed within the plot for wildlife surveillance. Over the past years, the recorded animal species captured by the cameras include: mammalian species such as Formosan Reeves's muntjac, Formosan leopard cat, Formosan rock macaque, Formosan pangolin, masked palm civet (locally known as white-nosed coati), red-bellied tree squirrel, Formosan wild boar, and crab-eating mongoose; avian species including Swinhoe's blue pheasant, Chinese bamboo partridge, Taiwan bamboo partridge, Malayan night heron, tiger thrush, and common emerald dove. The data collection protocol is as follows: the camera will record for 30 seconds immediately after being triggered by a passing animal, then stop recording. A 5-minute recording interval is set to avoid repeated recordings triggered by the same animal moving back and forth, which would cause excessive file storage, insufficient memory card capacity, and rapid battery depletion leading to power loss. Currently, data retrieval is carried out once per month: field personnel visit the plot to replace memory cards and recharge batteries, then collect the image files. These files are manually archived by camera ID, with filenames named based on the image capture date. Subsequently, each individual image file is manually reviewed to conduct animal classification and identification. The classified data is organized by animal species, with metrics including daily detection counts, monthly detection counts, and annual total counts, which serve as datasets for future analytical studies.