{"componentChunkName":"component---src-templates-acg-portal-new-template-tsx","path":"/umu0pb62c","result":{"data":{"markdownRemark":{"html":"<h2 id=\"简介\"><a href=\"#%E7%AE%80%E4%BB%8B\" aria-label=\"简介 permalink\" class=\"anchor\"><svg aria-hidden=\"true\" focusable=\"false\" height=\"16\" version=\"1.1\" viewBox=\"0 0 16 16\" width=\"16\"><path fill-rule=\"evenodd\" d=\"M4 9h1v1H4c-1.5 0-3-1.69-3-3.5S2.55 3 4 3h4c1.45 0 3 1.69 3 3.5 0 1.41-.91 2.72-2 3.25V8.59c.58-.45 1-1.27 1-2.09C10 5.22 8.98 4 8 4H4c-.98 0-2 1.22-2 2.5S3 9 4 9zm9-3h-1v1h1c1 0 2 1.22 2 2.5S13.98 12 13 12H9c-.98 0-2-1.22-2-2.5 0-.83.42-1.64 1-2.09V6.25c-1.09.53-2 1.84-2 3.25C6 11.31 7.55 13 9 13h4c1.45 0 3-1.69 3-3.5S14.5 6 13 6z\"></path></svg></a>简介</h2>\n<p>视频抽帧打标算子：按 <code>all_keyframes</code> 或 <code>uniform</code> 抽帧，逐帧过 RAM++（recognize-anything plus）打标，再按标签出现频次降序合并成一条视频级标签列表。</p>\n<h3 id=\"功能描述\"><a href=\"#%E5%8A%9F%E8%83%BD%E6%8F%8F%E8%BF%B0\" aria-label=\"功能描述 permalink\" class=\"anchor\"><svg aria-hidden=\"true\" focusable=\"false\" height=\"16\" version=\"1.1\" viewBox=\"0 0 16 16\" width=\"16\"><path fill-rule=\"evenodd\" d=\"M4 9h1v1H4c-1.5 0-3-1.69-3-3.5S2.55 3 4 3h4c1.45 0 3 1.69 3 3.5 0 1.41-.91 2.72-2 3.25V8.59c.58-.45 1-1.27 1-2.09C10 5.22 8.98 4 8 4H4c-.98 0-2 1.22-2 2.5S3 9 4 9zm9-3h-1v1h1c1 0 2 1.22 2 2.5S13.98 12 13 12H9c-.98 0-2-1.22-2-2.5 0-.83.42-1.64 1-2.09V6.25c-1.09.53-2 1.84-2 3.25C6 11.31 7.55 13 9 13h4c1.45 0 3-1.69 3-3.5S14.5 6 13 6z\"></path></svg></a>功能描述</h3>\n<ul>\n<li>抽帧语义：<code>all_keyframes</code> 只解关键帧（I 帧，PyAV 的 <code>skip_frame=\"NONKEY\"</code>）；<code>uniform</code> 下 <code>frame_num=1</code> 取中间帧、<code>=2</code> 取首尾帧、<code>>2</code> 在时长内均匀取（含首尾）</li>\n<li><code>max_frames</code>（默认 64）是显存兜底：抽出的帧超过上限时再按同一均匀规则抽稀，否则长视频关键帧过多会打爆显存</li>\n<li>打标按 <code>batch_size</code> 分批，且整列的 <code>(行号, 帧)</code> 会先展平再统一分批，一个 batch 可以跨行，避免逐视频起停模型</li>\n<li>帧级标签先在帧内去重（保序），再跨帧计频；因此最终频次等于「出现该标签的帧数」，同一帧内重复出现的标签不重复计数</li>\n<li>视频级输出按频次降序、已去重（同频次保持首次出现顺序）</li>\n<li>权重只从本地 <code>&#x3C;model_path>/&#x3C;model_name></code> 加载，文件缺失时 <code>__init__</code> 抛 <code>FileNotFoundError</code>，不联网、不做运行时 pip 安装</li>\n<li>兼容处理：RAM++ 的 BERT 代码从 <code>transformers.modeling_utils</code> 导三个已迁走的函数，算子加载前会从 <code>transformers.pytorch_utils</code> 回填同名别名，不需要为它降级整个环境的 transformers</li>\n<li>解码走 PyAV 而不是 torchcodec：torchcodec 依赖 CUDA NPP 动态库（<code>libnppicc.so</code>），镜像缺库时整个模块 import 就会失败</li>\n<li>单行抽帧失败或某个 batch 推理失败只记日志，对应行返回空列表，不影响同批其它行</li>\n</ul>\n<h2 id=\"算子参数\"><a href=\"#%E7%AE%97%E5%AD%90%E5%8F%82%E6%95%B0\" aria-label=\"算子参数 permalink\" class=\"anchor\"><svg aria-hidden=\"true\" focusable=\"false\" height=\"16\" version=\"1.1\" viewBox=\"0 0 16 16\" width=\"16\"><path fill-rule=\"evenodd\" d=\"M4 9h1v1H4c-1.5 0-3-1.69-3-3.5S2.55 3 4 3h4c1.45 0 3 1.69 3 3.5 0 1.41-.91 2.72-2 3.25V8.59c.58-.45 1-1.27 1-2.09C10 5.22 8.98 4 8 4H4c-.98 0-2 1.22-2 2.5S3 9 4 9zm9-3h-1v1h1c1 0 2 1.22 2 2.5S13.98 12 13 12H9c-.98 0-2-1.22-2-2.5 0-.83.42-1.64 1-2.09V6.25c-1.09.53-2 1.84-2 3.25C6 11.31 7.55 13 9 13h4c1.45 0 3-1.69 3-3.5S14.5 6 13 6z\"></path></svg></a>算子参数</h2>\n<h3 id=\"输入\"><a href=\"#%E8%BE%93%E5%85%A5\" aria-label=\"输入 permalink\" class=\"anchor\"><svg aria-hidden=\"true\" focusable=\"false\" height=\"16\" version=\"1.1\" viewBox=\"0 0 16 16\" width=\"16\"><path fill-rule=\"evenodd\" d=\"M4 9h1v1H4c-1.5 0-3-1.69-3-3.5S2.55 3 4 3h4c1.45 0 3 1.69 3 3.5 0 1.41-.91 2.72-2 3.25V8.59c.58-.45 1-1.27 1-2.09C10 5.22 8.98 4 8 4H4c-.98 0-2 1.22-2 2.5S3 9 4 9zm9-3h-1v1h1c1 0 2 1.22 2 2.5S13.98 12 13 12H9c-.98 0-2-1.22-2-2.5 0-.83.42-1.64 1-2.09V6.25c-1.09.53-2 1.84-2 3.25C6 11.31 7.55 13 9 13h4c1.45 0 3-1.69 3-3.5S14.5 6 13 6z\"></path></svg></a>输入</h3>\n<table>\n<thead>\n<tr>\n<th>输入</th>\n<th>含义</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>video</td>\n<td>视频输入数组，内容类型由 video_src_type 决定（本地/BOS 路径 / Base64 字符串 / bytes）</td>\n</tr>\n</tbody>\n</table>\n<h3 id=\"输出\"><a href=\"#%E8%BE%93%E5%87%BA\" aria-label=\"输出 permalink\" class=\"anchor\"><svg aria-hidden=\"true\" focusable=\"false\" height=\"16\" version=\"1.1\" viewBox=\"0 0 16 16\" width=\"16\"><path fill-rule=\"evenodd\" d=\"M4 9h1v1H4c-1.5 0-3-1.69-3-3.5S2.55 3 4 3h4c1.45 0 3 1.69 3 3.5 0 1.41-.91 2.72-2 3.25V8.59c.58-.45 1-1.27 1-2.09C10 5.22 8.98 4 8 4H4c-.98 0-2 1.22-2 2.5S3 9 4 9zm9-3h-1v1h1c1 0 2 1.22 2 2.5S13.98 12 13 12H9c-.98 0-2-1.22-2-2.5 0-.83.42-1.64 1-2.09V6.25c-1.09.53-2 1.84-2 3.25C6 11.31 7.55 13 9 13h4c1.45 0 3-1.69 3-3.5S14.5 6 13 6z\"></path></svg></a>输出</h3>\n<table>\n<thead>\n<tr>\n<th>输出</th>\n<th>含义</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>frame_tags</td>\n<td>list&#x3C;large_string>，视频级标签，按出现频次降序去重；输入为 null 或该行失败时为空列表</td>\n</tr>\n</tbody>\n</table>\n<h3 id=\"参数\"><a href=\"#%E5%8F%82%E6%95%B0\" aria-label=\"参数 permalink\" class=\"anchor\"><svg aria-hidden=\"true\" focusable=\"false\" height=\"16\" version=\"1.1\" viewBox=\"0 0 16 16\" width=\"16\"><path fill-rule=\"evenodd\" d=\"M4 9h1v1H4c-1.5 0-3-1.69-3-3.5S2.55 3 4 3h4c1.45 0 3 1.69 3 3.5 0 1.41-.91 2.72-2 3.25V8.59c.58-.45 1-1.27 1-2.09C10 5.22 8.98 4 8 4H4c-.98 0-2 1.22-2 2.5S3 9 4 9zm9-3h-1v1h1c1 0 2 1.22 2 2.5S13.98 12 13 12H9c-.98 0-2-1.22-2-2.5 0-.83.42-1.64 1-2.09V6.25c-1.09.53-2 1.84-2 3.25C6 11.31 7.55 13 9 13h4c1.45 0 3-1.69 3-3.5S14.5 6 13 6z\"></path></svg></a>参数</h3>\n<table>\n<thead>\n<tr>\n<th>参数名称</th>\n<th>类型</th>\n<th>默认值</th>\n<th>描述</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>video_src_type</td>\n<td>str</td>\n<td>\"video_url\"</td>\n<td>输入形式：video_url / video_base64 / video_binary</td>\n</tr>\n<tr>\n<td>model_path</td>\n<td>str</td>\n<td>\"/opt/aihc/model\"</td>\n<td>权重根目录</td>\n</tr>\n<tr>\n<td>model_name</td>\n<td>str</td>\n<td>\"ram_plus_swin_large_14m.pth\"</td>\n<td>相对 model_path 的 RAM++ 权重文件名</td>\n</tr>\n<tr>\n<td>frame_sampling_method</td>\n<td>str</td>\n<td>\"all_keyframes\"</td>\n<td>抽帧方式：all_keyframes / uniform</td>\n</tr>\n<tr>\n<td>frame_num</td>\n<td>int</td>\n<td>3</td>\n<td>uniform 模式抽帧数，必须为正</td>\n</tr>\n<tr>\n<td>input_size</td>\n<td>int</td>\n<td>384</td>\n<td>模型输入分辨率</td>\n</tr>\n<tr>\n<td>max_frames</td>\n<td>int</td>\n<td>64</td>\n<td>单视频抽帧上限，&#x3C;= 0 表示不限</td>\n</tr>\n<tr>\n<td>batch_size</td>\n<td>int</td>\n<td>8</td>\n<td>打标微批大小，&#x3C;= 1 时按 1 处理</td>\n</tr>\n<tr>\n<td>rank</td>\n<td>int</td>\n<td>0</td>\n<td>多卡场景 worker 序号，设备取 cuda:&#x3C;rank % 可见卡数></td>\n</tr>\n</tbody>\n</table>\n<h2 id=\"注意事项\"><a href=\"#%E6%B3%A8%E6%84%8F%E4%BA%8B%E9%A1%B9\" aria-label=\"注意事项 permalink\" class=\"anchor\"><svg aria-hidden=\"true\" focusable=\"false\" height=\"16\" version=\"1.1\" viewBox=\"0 0 16 16\" width=\"16\"><path fill-rule=\"evenodd\" d=\"M4 9h1v1H4c-1.5 0-3-1.69-3-3.5S2.55 3 4 3h4c1.45 0 3 1.69 3 3.5 0 1.41-.91 2.72-2 3.25V8.59c.58-.45 1-1.27 1-2.09C10 5.22 8.98 4 8 4H4c-.98 0-2 1.22-2 2.5S3 9 4 9zm9-3h-1v1h1c1 0 2 1.22 2 2.5S13.98 12 13 12H9c-.98 0-2-1.22-2-2.5 0-.83.42-1.64 1-2.09V6.25c-1.09.53-2 1.84-2 3.25C6 11.31 7.55 13 9 13h4c1.45 0 3-1.69 3-3.5S14.5 6 13 6z\"></path></svg></a>注意事项</h2>\n<ul>\n<li>标签频次是帧内去重后的计频（等于「出现该标签的帧数」），同频次标签按首次出现顺序排列，因此同频次标签之间的先后不代表置信度差异。</li>\n<li><code>all_keyframes</code> 抽出的帧数取决于视频的 GOP 结构，不可控；需要固定帧数就用 <code>uniform</code> + <code>frame_num</code>。</li>\n<li>backbone 固定为 <code>swin_l</code>（与默认权重匹配），换权重要确认结构一致；<code>input_size</code> 需与权重训练分辨率匹配。</li>\n<li><code>ram</code>（recognize-anything）包是运行依赖，算子不会在运行时安装它，缺失时 import 直接失败。</li>\n<li>是否用 GPU 由 <code>aihc_udf</code> 的 <code>num_gpus</code> 推导（基类按 <code>num_gpus > 0</code> 置 <code>use_gpu</code>）。</li>\n</ul>\n<h2 id=\"调用示例\"><a href=\"#%E8%B0%83%E7%94%A8%E7%A4%BA%E4%BE%8B\" aria-label=\"调用示例 permalink\" class=\"anchor\"><svg aria-hidden=\"true\" focusable=\"false\" height=\"16\" version=\"1.1\" viewBox=\"0 0 16 16\" width=\"16\"><path fill-rule=\"evenodd\" d=\"M4 9h1v1H4c-1.5 0-3-1.69-3-3.5S2.55 3 4 3h4c1.45 0 3 1.69 3 3.5 0 1.41-.91 2.72-2 3.25V8.59c.58-.45 1-1.27 1-2.09C10 5.22 8.98 4 8 4H4c-.98 0-2 1.22-2 2.5S3 9 4 9zm9-3h-1v1h1c1 0 2 1.22 2 2.5S13.98 12 13 12H9c-.98 0-2-1.22-2-2.5 0-.83.42-1.64 1-2.09V6.25c-1.09.53-2 1.84-2 3.25C6 11.31 7.55 13 9 13h4c1.45 0 3-1.69 3-3.5S14.5 6 13 6z\"></path></svg></a>调用示例</h2>\n\n    <div class=\"code-block-wrapper\">\n        <div class=\"code-block\">\n            <div class=\"code-block-header\">\n                <span class=\"code-block-name\">Python</span>\n                <button class=\"code-copy-btn\" data-tooltip-text=\"\">\n                    <svg xmlns=\"http://www.w3.org/2000/svg\" width=\"16\" height=\"16\" viewBox=\"0 0 16 16\" fill=\"none\"> <path fill-rule=\"evenodd\" clip-rule=\"evenodd\" d=\"M5.57894 3.45614C5.57894 3.38832 5.63392 3.33333 5.70175 3.33333H12.5439C12.6117 3.33333 12.6667 3.38832 12.6667 3.45614V10.2982C12.6667 10.3661 12.6117 10.4211 12.5439 10.4211H11.7544V5.70175C11.7544 4.89754 11.1025 4.24561 10.2982 4.24561H5.57894V3.45614ZM4.24561 4.24561V3.45614C4.24561 2.65194 4.89754 2 5.70175 2H12.5439C13.3481 2 14 2.65194 14 3.45614V10.2982C14 11.1025 13.3481 11.7544 12.5439 11.7544H11.7544V12.5439C11.7544 13.3481 11.1025 14 10.2982 14H3.45614C2.65194 14 2 13.3481 2 12.5439V5.70175C2 4.89754 2.65194 4.24561 3.45614 4.24561H4.24561ZM3.33333 5.70175C3.33333 5.63392 3.38832 5.57894 3.45614 5.57894H10.2982C10.3661 5.57894 10.4211 5.63392 10.4211 5.70175V12.5439C10.4211 12.6117 10.3661 12.6667 10.2982 12.6667H3.45614C3.38832 12.6667 3.33333 12.6117 3.33333 12.5439V5.70175Z\" fill=\"currentColor\"></path> </svg>\n                    复制\n                </button>\n            </div>\n            <div class=\"code-block-content\">\n                <pre class=\"language-python\"><code><span class=\"line-number\">1</span><span class=\"token keyword\">from</span> __future__ <span class=\"token keyword\">import</span> annotations\n<span class=\"line-number\">2</span>\n<span class=\"line-number\">3</span><span class=\"token keyword\">import</span> os\n<span class=\"line-number\">4</span>\n<span class=\"line-number\">5</span><span class=\"token keyword\">import</span> daft\n<span class=\"line-number\">6</span><span class=\"token keyword\">from</span> daft <span class=\"token keyword\">import</span> col\n<span class=\"line-number\">7</span>\n<span class=\"line-number\">8</span><span class=\"token keyword\">from</span> daft<span class=\"token punctuation\">.</span>aihc<span class=\"token punctuation\">.</span>common<span class=\"token punctuation\">.</span>udf <span class=\"token keyword\">import</span> aihc_udf\n<span class=\"line-number\">9</span><span class=\"token keyword\">from</span> daft<span class=\"token punctuation\">.</span>aihc<span class=\"token punctuation\">.</span>functions<span class=\"token punctuation\">.</span>video<span class=\"token punctuation\">.</span>video_tagging_from_frames <span class=\"token keyword\">import</span> VideoTaggingFromFrames\n<span class=\"line-number\">10</span>\n<span class=\"line-number\">11</span>os<span class=\"token punctuation\">.</span>environ<span class=\"token punctuation\">.</span>setdefault<span class=\"token punctuation\">(</span><span class=\"token string\">\"BOS_ENDPOINT\"</span><span class=\"token punctuation\">,</span> <span class=\"token string\">\"http://bj.bcebos.com\"</span><span class=\"token punctuation\">)</span>\n<span class=\"line-number\">12</span>os<span class=\"token punctuation\">.</span>environ<span class=\"token punctuation\">.</span>setdefault<span class=\"token punctuation\">(</span><span class=\"token string\">\"BOS_REGION\"</span><span class=\"token punctuation\">,</span> <span class=\"token string\">\"bj\"</span><span class=\"token punctuation\">)</span>\n<span class=\"line-number\">13</span>\n<span class=\"line-number\">14</span><span class=\"token keyword\">if</span> __name__ <span class=\"token operator\">==</span> <span class=\"token string\">\"__main__\"</span><span class=\"token punctuation\">:</span>\n<span class=\"line-number\">15</span>    <span class=\"token keyword\">if</span> os<span class=\"token punctuation\">.</span>getenv<span class=\"token punctuation\">(</span><span class=\"token string\">\"DAFT_RUNNER\"</span><span class=\"token punctuation\">,</span> <span class=\"token string\">\"native\"</span><span class=\"token punctuation\">)</span> <span class=\"token operator\">==</span> <span class=\"token string\">\"ray\"</span><span class=\"token punctuation\">:</span>\n<span class=\"line-number\">16</span>        <span class=\"token keyword\">import</span> ray\n<span class=\"line-number\">17</span>        ray<span class=\"token punctuation\">.</span>init<span class=\"token punctuation\">(</span>dashboard_host<span class=\"token operator\">=</span><span class=\"token string\">\"0.0.0.0\"</span><span class=\"token punctuation\">,</span> ignore_reinit_error<span class=\"token operator\">=</span><span class=\"token boolean\">True</span><span class=\"token punctuation\">)</span>\n<span class=\"line-number\">18</span>        daft<span class=\"token punctuation\">.</span>set_runner_ray<span class=\"token punctuation\">(</span><span class=\"token punctuation\">)</span>\n<span class=\"line-number\">19</span>    daft<span class=\"token punctuation\">.</span>set_execution_config<span class=\"token punctuation\">(</span>actor_udf_ready_timeout<span class=\"token operator\">=</span><span class=\"token number\">6000</span><span class=\"token punctuation\">,</span> min_cpu_per_task<span class=\"token operator\">=</span><span class=\"token number\">0</span><span class=\"token punctuation\">)</span>\n<span class=\"line-number\">20</span>\n<span class=\"line-number\">21</span>ds <span class=\"token operator\">=</span> daft<span class=\"token punctuation\">.</span>from_pydict<span class=\"token punctuation\">(</span><span class=\"token punctuation\">{</span><span class=\"token string\">\"video\"</span><span class=\"token punctuation\">:</span> <span class=\"token punctuation\">[</span><span class=\"token string\">\"bos://your-bucket/sample.mp4\"</span><span class=\"token punctuation\">]</span><span class=\"token punctuation\">}</span><span class=\"token punctuation\">)</span>\n<span class=\"line-number\">22</span>    ds <span class=\"token operator\">=</span> ds<span class=\"token punctuation\">.</span>with_column<span class=\"token punctuation\">(</span>\n<span class=\"line-number\">23</span>        <span class=\"token string\">\"frame_tags\"</span><span class=\"token punctuation\">,</span>\n<span class=\"line-number\">24</span>        aihc_udf<span class=\"token punctuation\">(</span>\n<span class=\"line-number\">25</span>            VideoTaggingFromFrames<span class=\"token punctuation\">,</span>\n<span class=\"line-number\">26</span>            construct_args<span class=\"token operator\">=</span><span class=\"token punctuation\">{</span>\n<span class=\"line-number\">27</span>                <span class=\"token string\">\"video_src_type\"</span><span class=\"token punctuation\">:</span> <span class=\"token string\">\"video_url\"</span><span class=\"token punctuation\">,</span>\n<span class=\"line-number\">28</span>                <span class=\"token string\">\"model_path\"</span><span class=\"token punctuation\">:</span> <span class=\"token string\">\"/path/to/models\"</span><span class=\"token punctuation\">,</span>\n<span class=\"line-number\">29</span>                <span class=\"token string\">\"model_name\"</span><span class=\"token punctuation\">:</span> <span class=\"token string\">\"ram_plus_swin_large_14m.pth\"</span><span class=\"token punctuation\">,</span>\n<span class=\"line-number\">30</span>                <span class=\"token string\">\"frame_sampling_method\"</span><span class=\"token punctuation\">:</span> <span class=\"token string\">\"uniform\"</span><span class=\"token punctuation\">,</span>\n<span class=\"line-number\">31</span>                <span class=\"token string\">\"frame_num\"</span><span class=\"token punctuation\">:</span> <span class=\"token number\">3</span><span class=\"token punctuation\">,</span>\n<span class=\"line-number\">32</span>                <span class=\"token string\">\"batch_size\"</span><span class=\"token punctuation\">:</span> <span class=\"token number\">4</span><span class=\"token punctuation\">,</span>\n<span class=\"line-number\">33</span>            <span class=\"token punctuation\">}</span><span class=\"token punctuation\">,</span>\n<span class=\"line-number\">34</span>            num_cpus<span class=\"token operator\">=</span><span class=\"token number\">1</span><span class=\"token punctuation\">,</span>\n<span class=\"line-number\">35</span>            num_gpus<span class=\"token operator\">=</span><span class=\"token number\">1</span><span class=\"token punctuation\">,</span>\n<span class=\"line-number\">36</span>            concurrency<span class=\"token operator\">=</span><span class=\"token number\">1</span><span class=\"token punctuation\">,</span>\n<span class=\"line-number\">37</span>            batch_size<span class=\"token operator\">=</span><span class=\"token number\">1</span><span class=\"token punctuation\">,</span>\n<span class=\"line-number\">38</span>        <span class=\"token punctuation\">)</span><span class=\"token punctuation\">(</span>col<span class=\"token punctuation\">(</span><span class=\"token string\">\"video\"</span><span class=\"token punctuation\">)</span><span class=\"token punctuation\">)</span><span class=\"token punctuation\">,</span>\n<span class=\"line-number\">39</span>    <span class=\"token punctuation\">)</span>\n<span class=\"line-number\">40</span>    ds<span class=\"token punctuation\">.</span>show<span class=\"token punctuation\">(</span><span class=\"token punctuation\">)</span></code></pre>\n            </div>\n        </div>\n    </div>\n  ","fields":{"slug":"umu0pb62c","title":"视频帧标签","date":"2026-09-14","extractedHeadings":[]},"headings":[{"value":"简介","depth":2},{"value":"功能描述","depth":3},{"value":"算子参数","depth":2},{"value":"输入","depth":3},{"value":"输出","depth":3},{"value":"参数","depth":3},{"value":"注意事项","depth":2},{"value":"调用示例","depth":2}]}},"pageContext":{"isCreatedByStatefulCreatePages":false,"slug":"umu0pb62c","prev":{"id":"xmu0pfbjc","name":"视频目标分割","path":"xmu0pfbjc","filePath":"操作指南/AI数据处理/算子列表/视频/视频目标分割.md","seo":null,"parentIds":["ilib2qygp","Ymo88m8hi","Imob3m6so","6mpawyry0"],"parents":[{"id":"ilib2qygp","documentId":"bfa43a8b-968a-41a1-8c9d-906507eeaed9","name":"操作指南","repoName":"AIHC","filePath":"操作指南","disabled":false,"path":"ilib2qygp","lastMergeTime":null,"isApiDoc":null,"httpMethod":null,"seo":null,"sourceOrgName":null,"sourceRepoName":null,"sourceDocumentId":null,"solutionTag":null,"solutionColor":null},{"id":"Ymo88m8hi","documentId":"c8cb5e38-f8c5-40f4-a424-b0c7895f0c0a","name":"AI数据处理","repoName":"AIHC","filePath":"操作指南/AI数据处理","disabled":false,"path":"Ymo88m8hi","lastMergeTime":"2026-04-21 14:23:10","isApiDoc":null,"httpMethod":null,"seo":null,"sourceOrgName":null,"sourceRepoName":null,"sourceDocumentId":"","solutionTag":null,"solutionColor":null},{"id":"Imob3m6so","documentId":"fe548e34-6659-4ff5-86f6-eee2c43aec90","name":"算子列表","repoName":"AIHC","filePath":"操作指南/AI数据处理/算子列表","disabled":false,"path":"Imob3m6so","lastMergeTime":null,"isApiDoc":null,"httpMethod":null,"seo":null,"sourceOrgName":null,"sourceRepoName":null,"sourceDocumentId":"","solutionTag":null,"solutionColor":null},{"id":"6mpawyry0","documentId":"3428d02c-e287-45b8-9d16-5a7b20881dd4","name":"视频","repoName":"AIHC","filePath":"操作指南/AI数据处理/算子列表/视频","disabled":false,"path":"6mpawyry0","lastMergeTime":null,"isApiDoc":null,"httpMethod":null,"seo":null,"sourceOrgName":null,"sourceRepoName":null,"sourceDocumentId":"","solutionTag":null,"solutionColor":null}]},"next":{"id":"smohxoure","name":"AIHC-Daft算子开发使用指南","path":"smohxoure","filePath":"操作指南/AI数据处理/AIHC-Daft算子开发使用指南.md","seo":null,"parentIds":["ilib2qygp","Ymo88m8hi"],"parents":[{"id":"ilib2qygp","documentId":"bfa43a8b-968a-41a1-8c9d-906507eeaed9","name":"操作指南","repoName":"AIHC","filePath":"操作指南","disabled":false,"path":"ilib2qygp","lastMergeTime":null,"isApiDoc":null,"httpMethod":null,"seo":null,"sourceOrgName":null,"sourceRepoName":null,"sourceDocumentId":null,"solutionTag":null,"solutionColor":null},{"id":"Ymo88m8hi","documentId":"c8cb5e38-f8c5-40f4-a424-b0c7895f0c0a","name":"AI数据处理","repoName":"AIHC","filePath":"操作指南/AI数据处理","disabled":false,"path":"Ymo88m8hi","lastMergeTime":"2026-04-21 14:23:10","isApiDoc":null,"httpMethod":null,"seo":null,"sourceOrgName":null,"sourceRepoName":null,"sourceDocumentId":"","solutionTag":null,"solutionColor":null}]},"parents":[{"id":"ilib2qygp","documentId":"bfa43a8b-968a-41a1-8c9d-906507eeaed9","name":"操作指南","repoName":"AIHC","filePath":"操作指南","disabled":false,"path":"ilib2qygp","lastMergeTime":null,"isApiDoc":null,"httpMethod":null,"seo":null,"sourceOrgName":null,"sourceRepoName":null,"sourceDocumentId":null,"solutionTag":null,"solutionColor":null},{"id":"Ymo88m8hi","documentId":"c8cb5e38-f8c5-40f4-a424-b0c7895f0c0a","name":"AI数据处理","repoName":"AIHC","filePath":"操作指南/AI数据处理","disabled":false,"path":"Ymo88m8hi","lastMergeTime":"2026-04-21 14:23:10","isApiDoc":null,"httpMethod":null,"seo":null,"sourceOrgName":null,"sourceRepoName":null,"sourceDocumentId":"","solutionTag":null,"solutionColor":null},{"id":"Imob3m6so","documentId":"fe548e34-6659-4ff5-86f6-eee2c43aec90","name":"算子列表","repoName":"AIHC","filePath":"操作指南/AI数据处理/算子列表","disabled":false,"path":"Imob3m6so","lastMergeTime":null,"isApiDoc":null,"httpMethod":null,"seo":null,"sourceOrgName":null,"sourceRepoName":null,"sourceDocumentId":"","solutionTag":null,"solutionColor":null},{"id":"6mpawyry0","documentId":"3428d02c-e287-45b8-9d16-5a7b20881dd4","name":"视频","repoName":"AIHC","filePath":"操作指南/AI数据处理/算子列表/视频","disabled":false,"path":"6mpawyry0","lastMergeTime":null,"isApiDoc":null,"httpMethod":null,"seo":null,"sourceOrgName":null,"sourceRepoName":null,"sourceDocumentId":"","solutionTag":null,"solutionColor":null}],"specificSeo":null}}}