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。

Currently, at the Ecological Sensing Platform of the Taiwan Forestry Research Institute (TFRI) at http://iesn.tfri.gov.tw/forestDW/, a 25-hectare (500 m × 500 m) long-term dynamic broad-leaved forest plot has been established between Weir No.1 and Weir No.2 of the TFRI's experimental forest at the Lianhuachi Research Center, for three monitoring purposes: anuran acoustic monitoring, online live streaming monitoring of the rare parasitic plant *Mitrastemon formosanum* (diamond-shaped nutgrass), and wildlife monitoring. Infrared camera traps have been deployed within the plot for wildlife monitoring.

Over the past years, the camera traps have captured footage of the following animal species: 8 mammalian species including Reeves' muntjac (*Muntiacus reevesi*), leopard cat (*Prionailurus bengalensis*), Formosan rock macaque (*Macaca cyclopis*), Chinese pangolin (*Manis pentadactyla*), masked palm civet (*Paguma larvata*), red-bellied squirrel (*Callosciurus erythraeus*), wild boar (*Sus scrofa*), and crab-eating mongoose (*Herpestes urva*); and 6 avian species including Swinhoe's blue pheasant (*Lophura swinhoii*), Chinese bamboo partridge (*Bambusicola thoracicus*), mountain bamboo partridge (*Bambusicola fytchii*), Malayan night heron (*Gorsachius melanolophus*), tiger thrush (*Zoothera tigris*), and common emerald dove (*Chalcophaps indica*).

The recording configuration is set as follows: the camera will cease recording 30 seconds after being triggered by an animal, with a 5-minute interval between consecutive triggering events. This setting prevents excessive file accumulation, insufficient memory card capacity, and rapid battery depletion caused by repeated triggering from animals moving back and forth within the detection range.

Currently, field personnel retrieve the image files once per month by visiting the plot to replace memory cards and recharge camera batteries. The retrieved footage is manually archived by camera ID, with filenames generated based on the image capture dates. Subsequently, each individual image file is manually reviewed to classify and identify the recorded animals. The classified data is then organized by animal species, with metrics including daily occurrence count, monthly occurrence count, and annual total count, which serve as datasets for subsequent analysis.