中EN
County
Colour by 淨寬Clear width 路緣斜坡Curb ramps
Obstacles ⚡ 電線桿⚡ Utility poles
Crashes 🔴 A1 事故點🔴 A1 crash points
🏫 School zones 國小Elementary 國中Junior high 高中職Senior/voc. 大專Tertiary 特教Special ed. 其他Other 依層級by level 依步行不足by walk-deficit Walkshed 🧑 青少年Teen 👦🧓 孩童・長者Child & elder ♿ 輪椅Wheelchair
Advocacy 🌳 300m 生活街道圈300 m living streets 🛣 生活街道候選 ≤12 mLiving-street candidates ≤12 m +12–15 m 邊界+12–15 m borderline 🧭 道路層級 Sustainable SafetyRoad hierarchy (Sustainable Safety)
—
— ↗ 看最新街景(Google Maps)Latest imagery (Google Maps)
載入街景中…Loading Street View…

2026 年 6 月路網(實際多邊形範圍,幾何經 ±0.3 公尺簡化)。放大後滑過任一段可看路名、寬度、淨寬與斜坡數;點擊路段即開啟街景比對;放大至街區尺度後,點擊任何沒有人行道資料的位置,也會直接開啟該點街景——用來檢查無人行道街道的實際路況;街景為主畫面、地圖縮至右上角,小地圖上的小人與視角錐標示實際全景位置與朝向:在全景內拖轉、前進時會即時跟動;「↻ 旋轉 90°」可快速換邊看。資訊列顯示影像拍攝日期;街景取該點最新一輪官方拍攝(若 Google 服務無法載入,會自動退回精簡嵌入模式,該模式下視角錐僅跟隨旋轉鈕)。資料依縣市按需載入(0.03–6.2 MB/縣市)。June 2026 network (actual polygon extents, geometry simplified within ±0.3 m). Zoom in and hover a segment for road name, width, clear width and ramp count; click a segment to open Street View comparison; at street zoom, clicking anywhere without a sidewalk segment opens Street View at that point — for inspecting roads that have no sidewalk — the panorama takes the stage and the map docks top-right, with a pegman + view cone marking the true panorama position and facing — they follow live as you drag or walk inside the panorama; "↻ Rotate 90°" flips sides quickly. The info bar shows the imagery capture date, and the panorama uses the newest official pass at that spot (if Google's service fails to load, a lightweight embed takes over, where the cone follows the rotate button only). Data loads on demand per county (0.03–6.2 MB each).

🔴 A1 事故點圖層:內政部警政署 A1(當場或 24 小時內死亡)交通事故(2018–2025 有座標者,全國 13,708 點,按縣市過濾顯示),點色依受害者運具(黑=機車、藍=汽/貨車、紅=行人、金=慢車)、點徑依死亡人數。把死亡事故疊在淨寬著色上,可直接檢視「行人死亡點位周邊的人行道是否存在、是否過得去」:紅點(行人)落在紅色路段(淨寬 <1.5 公尺)或根本沒有人行道的路口,就是最需優先改善的證據。滑過任一點可看日期、區、運具與死傷數。🔴 A1 crash layer: National Police Agency A1 fatal crashes (death on scene or within 24 h; 2018–2025 geocoded records, 13,708 points nationwide, filtered to the selected county). Dot colour = victim mode (black motorcycle, blue car/truck, red pedestrian, gold bicycle), dot size = deaths. Overlaying fatalities on the clear-width colours shows directly whether the sidewalks around each death exist and are passable — a red (pedestrian) dot on a red segment (<1.5 m clear) or on a corner with no sidewalk at all is priority-one evidence. Hover a dot for date, district, mode and casualties.

⚡ 電線桿圖層:台電桿籍資料(全國約 273 萬支,TWD97 座標轉換;本島 121 分帶、離島 119 分帶),開啟後放大至街區尺度(zoom ≥ 16)顯示視窗內桿位。電線桿常立於人行道上,是輪椅與嬰兒車的實體障礙:與淨寬著色對照可看出「帳面有人行道、實際過不去」的路段。注意:屏東縣桿籍近乎缺漏(僅 88 筆);桿位為台電登記位置,可能與現地略有出入。⚡ Utility-pole layer: Taipower pole registry (~2.73M poles nationwide, TWD97→WGS84; 121°E zone for the main island, 119°E for offshore counties). When on, poles in the viewport render at zoom ≥ 16. Poles standing on sidewalks are physical barriers for wheelchairs and strollers — overlay them on the clear-width colours to spot segments that exist on paper but cannot be passed. Note: Pingtung's registry is nearly empty (88 records); positions are as registered and may differ slightly on site.

🏫 學校範圍圖層:內政部國土測繪中心「各級學校範圍圖」(全國約 4,336 筆校地範圍,圖層代碼 121、115 年 4 月版;座標由 TWD97 TM2/EPSG:3826 轉為 WGS84)。學校層級由校名判讀並分成國小、國中、高中職、大專、特教、其他六類,各為可複選的獨立圖層並以不同顏色標示,疊在人行道路網下方——可看出哪些人行道落在學校周邊,也就是學童每天上下學實際會走的路段。與淨寬著色、行人事故點對照,能鎖定「校門口周邊卻沒有安全人行道」的優先改善路口。滑過任一範圍可看校名、層級與資料年月;點擊任一學校即畫出 15 分鐘步行圈——虛線圓=理想範圍、藍色斑塊=沿人行道實際可達範圍,斑塊越破碎、越空,代表該學區人行道越斷裂欠缺。可切換三種步行者:🧑 青少年(約 4 km/h,1 公里)、👦🧓 孩童・長者(約 3 km/h,750 公尺——實證研究顯示學童揹書包、等紅綠燈的實際步速與長者相近,交通工程設計亦將兩者歸為同一設計速度群)、♿ 輪椅/行動不便(約 2.4 km/h,600 公尺,且排除淨寬不足 0.9 公尺與「該設而未設」路緣斜坡的路段——不只走得慢,很多路段根本過不去)。🌳 300m 生活街道圈(倡議圖層):開啟後以每所學校為圓心畫出方圓 300 公尺綠色虛線圈——通學的最後一段路;點擊任一圈可放大並顯示倡議說明(行人優先、車速上限 20 km/h、穿越性車流退場),供簡報與政策倡議示意使用。🏫 School-zone layer: NLSC "school boundary map" (~4,336 school-land polygons nationwide, layer 121, April 2026 edition; reprojected TWD97 TM2 / EPSG:3826 → WGS84). School level is parsed from each name into six classes — elementary, junior high, senior/vocational, tertiary, special education and other — each an independent multi-select layer with its own colour, drawn beneath the sidewalk network to show which pavements sit within a school catchment (the streets children walk to and from school on). Cross-referenced with the clear-width colours and pedestrian crashes, it pinpoints priority corners where a school gate lacks a safe sidewalk nearby. Hover a zone for the school name, level and data month; click a school to draw its 15-min walkshed — dashed circle = ideal range, blue patches = what's actually reachable along sidewalks; the emptier and more fragmented the patches, the more broken the network. Three mobility profiles: 🧑 teen (~4 km/h, 1 km), 👦🧓 child & elder (~3 km/h, 750 m — studies show schoolchildren's effective walking speed, with backpacks and crossings, matches elders', and traffic engineering groups them at the same design speed), ♿ wheelchair (~2.4 km/h, 600 m, additionally excluding segments under 0.9 m clear width or missing required curb ramps — not just slower, many segments are simply impassable). 🌳 300 m living-streets ring (advocacy layer): draws a dashed green 300 m radius around every school — the last leg of the school run; click a ring to zoom in with an advocacy note (pedestrian priority, 20 km/h, no through-traffic) for presentations and policy work.

🛣 生活街道候選圖層(Taiwanwonnerf):以 OpenStreetMap 全台路網(2026 年 9 月)估算每一路段路寬,指認 15 公尺以下、可能成為生活街道的道路:≤8 公尺巷弄型、8–12 公尺社區街道,另可加開 12–15 公尺邊界案例。路寬優先取 OSM width 標籤(覆蓋率僅約 0.3%),其次依 lanes 推估(lanes×3.5+2),再依台灣路名慣例(弄 4 m、巷 6 m、街 10 m、路 12 m)與道路等級預設;點擊任一路段可見估算值與依據。全台 ≤12 公尺道路約 109,000 公里,佔路網長度八成以上。🧭 道路層級圖層:依荷蘭 Sustainable Safety(Duurzaam Veilig)三層功能分級——Through roads(國道、快速公路,黑線)、Distributor roads(省道、縣道、鄉道、市區幹道,藍線)、Access roads(其餘市區道路與巷弄,即生活街道候選)。荷蘭經驗是:功能、設計與使用三者一致,Access road 才能落實 30 km/h 與行人優先;台灣的 tertiary(鄉道、市區次要幹道)常兼具集散與出入功能,屬「灰色道路」,暫歸 Distributor。方法與資料:Taiwanwonnerf(含各縣市 Shapefile)。🛣 Living-street candidates (Taiwanwonnerf): every OSM road segment in Taiwan (Sept 2026) gets an estimated width; segments under 15 m are flagged as potential living streets — ≤8 m alleys and 8–12 m local streets, plus an optional 12–15 m borderline class. Width comes from the OSM width tag where present (~0.3% coverage), else lanes×3.5+2, else Taiwanese road-name conventions (弄 4 m, 巷 6 m, 街 10 m, 路 12 m) and highway-class defaults; click any segment to see the estimate and its basis. Roads ≤12 m total ~109,000 km, over 80% of network length. 🧭 Road hierarchy: the Dutch Sustainable Safety three-tier functional classification — Through roads (freeways/expressways, black), Distributor roads (provincial/county/township roads and urban arterials, blue), Access roads (all other local streets, i.e. the living-street candidates). Dutch practice keeps function, design and use consistent so that access roads can carry 30 km/h and pedestrian priority; Taiwan's tertiary roads often mix both roles ("grey roads") and are provisionally classed as distributors. Method and data: Taiwanwonnerf (with per-county shapefiles).

📸 街道改造相簿Street Makeover Album

在街景模式按「📸 收藏此景」,即以官方 Street View Static API 擷取當下視角(全景 ID+方位角),存進這本相簿——收集你想改造的街道空間。每筆可按「🎨 AI 改造」一鍵送往 AI 街景生成器(圖片網址自動複製、開啟工具即可貼上生成改造願景圖)。相簿存在這台瀏覽器(localStorage),可加註記、匯出清單。In Street View, press "📸 Save view" to capture the current viewpoint (pano ID + heading) via the official Street View Static API into this album — a collection of streets you want to remake. Each entry has a "🎨 AI remake" button that hands the view to the AI street designer (image URL auto-copied for pasting). Stored in this browser (localStorage); add notes and export the list.
⬇ 匯出 CSVExport CSV ⬇ 匯出 JSONExport JSON 🗑 清空相簿Clear all

🏫 15 分鐘沿人行道可達分析15-min sidewalk-network reach

以各校為起點,沿實際人行道網路計算 15 分鐘(步行約 1 km、單車約 3 km)可達範圍,量化人行道的斷裂與欠缺。不足分數越高=步行環境越差;「連通率」為可達人行道 ÷ 直線範圍內人行道(越低代表越破碎)。想看單一學校的可達範圍圖:到上方地圖開啟學校圖層後點擊該校即可。From each school, 15-minute reach is traced along the real sidewalk network (~1 km walk / ~3 km bike), quantifying fragmentation and gaps. Higher deficit score = worse; "connectivity" = reachable ÷ straight-line sidewalk (lower = more broken). To see one school's reach map, enable the school layer on the map above and click the school.
County
Level
Grade
—
方法備註:人行道由「面」段以 15 公尺鄰近門檻接成步行網路(寬幹道視為需穿越的障礙而不連通);長度以面積÷寬度估算。此為模型近似,用於相對比較與優先排序,非現地量測。離島(澎湖/金門/連江)不在本圖資範圍。已有街道路網資料的縣市另以真實街道網路計算,紅字「街道覆蓋」=步行圈內該有人行道的街道中有人行道者的長度比例、「缺口」=無人行道街道的公里數。嘉義縣採 OSM 路網(主次要道路與巷弄,不含農路,以 25 公尺鄰近判定);台北市採北市府官方 8 公尺以上道路圖資(含官方人行道面積欄位與路口拓撲,僅計 8 公尺以上道路)。兩者路網基準不同,跨縣市比較時請以量級為準。Method: sidewalk polygons are linked into a walking graph with a 15 m proximity threshold (wide arterials act as uncrossable breaks); segment length ≈ area ÷ width. A modelled approximation for relative ranking, not field survey. Offshore counties are outside this dataset. Counties with street-network data additionally get a real street-network audit — red "street cover" = share (by length) of sidewalk-worthy streets in the walk zone that have a sidewalk, "gap" = km without one. Chiayi uses the OSM network (arterials + lanes, no farm tracks, 25 m proximity test); Taipei uses the city government's official ≥8 m road inventory (official sidewalk-area field + junction topology; ≥8 m roads only). Baselines differ — compare across counties by magnitude.

資料來源:交通部「全國人行道整體區位」開放資料,兩次全國普查快照:2023 年 6 月(83,301 個路段,全國 10,167 公里)、2026 年 6 月(92,326 個路段,全國 10,939 公里)。逐路段欄位包含道路名稱、路段長度、人行道寬度、淨寬(扣除路燈、招牌等障礙物後的可通行寬度)、路緣斜坡設置數量、所在縣市與鄉鎮市區。Source: MOTC's national sidewalk-location open dataset, two nationwide snapshots — June 2023 (83,301 segments, 10,167 km total) and June 2026 (92,326 segments, 10,939 km total). Per-segment fields include road name, segment length, sidewalk width, clear width (net of obstructions such as poles and signage), curb-ramp count, and county/district.

路緣斜坡缺口率怎麼算:原始欄位 SW_RAMP 是「已設置之路緣斜坡數量」,但若此處本來就不需要斜坡(例如路段中段、無側向通行需求),會標記為「N」(2023)或留空(2026)。我們只在「需要斜坡」的路段中,計算 SW_RAMP 為 0(該裝而未裝)的比例,作為缺口率:這是一個保守、只算「真正缺口」的指標。How the ramp gap rate is computed: the source field SW_RAMP counts installed curb ramps, but is marked "N" (2023) or left blank (2026) where no ramp is needed at all. We compute the gap rate only among segments that DO need a ramp, as the share where SW_RAMP = 0 (needed but not installed) — a conservative measure of genuine gaps only.

已知限制:2023 年原始資料為 TWD97 二度分帶座標系統(EPSG:3826),2026 年官方已提供 WGS84 版本;本頁的彙整統計僅使用逐路段的「縣市/鄉鎮市區」欄位,未使用逐路段座標,故不受座標系統差異影響。高雄市那瑪夏區與連江縣東引鄉的人行道資料存在於原始資料中,均已計入縣市統計。嘉義縣資料中有 2 筆路段的縣市欄位誤植(應屬嘉義市西區),已於彙整時更正;苗栗縣「頭份市」與彰化縣「員林市」在原始資料中仍使用 2015 年改制前的舊名(頭份鎮/員林鎮),已於彙整時對應至現行名稱。Known limitations: the 2023 source used the TWD97 TM2 coordinate system (EPSG:3826); the 2026 release already ships a WGS84 version. This page's aggregate statistics use only the per-segment county/district fields, never per-segment coordinates, so they are unaffected by the coordinate-system difference. Sidewalk records for 那瑪夏 (Kaohsiung) and 東引鄉 (Lienchiang) are included in the county statistics. Two segments in the Chiayi County file were mistagged (they belong to Chiayi City's West District) and were corrected during aggregation; 頭份/員林 (upgraded from 鎮 to 市 in 2015) are mapped to their current names.

Sources · MOTC 全國人行道整體區位資料 (data.gov.tw) · MOI Household Registration dataset 8410 (population) · g0v/twgeojson district boundaries

資料夥伴 · 合作單位
In partnership with
TCAN 台灣氣候行動網路 · Vision Zero Taiwan 還路於民行人路權促進會
還路於民 Vision Zero Taiwan · 官方網站Vision Zero Taiwan · Official sitevisionzero.tw ↗