{"id":401,"date":"2025-02-20T10:48:01","date_gmt":"2025-02-20T10:48:01","guid":{"rendered":"https:\/\/boochlin.com\/?p=401"},"modified":"2025-06-10T08:07:20","modified_gmt":"2025-06-10T08:07:20","slug":"android-%e7%88%9b%e8%a3%9d%e7%bd%ae%e6%83%b3%e8%b7%91%e4%ba%ba%e8%87%89%e8%be%a8%e8%ad%98-%e7%b3%9f%e7%b3%95%e7%9a%84%e7%92%b0%e5%a2%83%e6%89%8d%e8%83%bd%e7%b7%b4%e7%bf%92%e6%95%88%e8%83%bd%e5%84%aa","status":"publish","type":"post","link":"https:\/\/boochlin.com\/?p=401","title":{"rendered":"Android \u721b\u88dd\u7f6e\u60f3\u8dd1\u4eba\u81c9\u8fa8\u8b580 – \u5230\u5e95\u6709\u591a\u721b\uff1f"},"content":{"rendered":"\n
\"\"<\/a><\/figure>\n\n\n\n

\u721b\u88dd\u7f6e\u662f\u4ec0\u9ebc\uff1f<\/h2>\n\n\n\n

\u9019\u6b21\u7684\u4e3b\u89d2\u662f\uff1aQualcomm Snapdragon QM215<\/strong>\uff0c\u642d\u914d 2GB RAM<\/strong>\uff0c\u67b6\u69cb\u70ba ARMv8-A<\/strong>\u3002<\/p>\n\n\n\n

\u9019\u9846 SoC \u662f\u9ad8\u901a\u5728 2019 \u5e74\u767c\u552e<\/strong>\u7684\u8d85\u5165\u9580\u8655\u7406\u5668\uff0c\u8a2d\u8a08\u521d\u8877\u5c31\u662f\u7d66\u4f4e\u50f9 Android \u624b\u6a5f\u6216 IoT \u8a2d\u5099\u4f7f\u7528\u3002\u96d6\u7136\u5b83\u652f\u63f4 64bit \u67b6\u69cb\u3001\u96d9\u93e1\u982d\u3001\u529f\u8017\u4f4e\uff0c
\u4f46\u540c\u6642\u4e5f\u6709\u5e7e\u500b\u786c\u50b7\uff1a<\/p>\n\n\n\n

    \n
  • \u6c92\u6709 NPU\uff0cAI \u8981\u9760 CPU\/GPU \u786c\u625b<\/li>\n\n\n\n
  • RAM \u4e0a\u9650 3GB\uff0c\u5be6\u969b\u53ea\u88dd\u4e86 2GB\uff08\u70ba\u4e86\u7701\u9322\uff09<\/li>\n\n\n\n
  • \u56db\u6838\u5fc3 Cortex-A53\uff0c\u6700\u9ad8\u53ea\u6709 1.3GHz<\/li>\n<\/ul>\n\n\n\n

    \u4f60\u662f\u516c\u53f8\u8acb\u4f86\u7684\u6280\u8853\u8ca0\u8cac\u4eba\uff0c2024 \u5e74\u6536\u5230\u9019\u9846\u6676\u7247\uff0c\u8981\u5728\u5b83\u4e0a\u9762\u505a\u81c9\u90e8\u8fa8\u8b58\u3002\u4f60\u4e0d\u80fd\u7ffb\u684c\uff0c\u4e5f\u4e0d\u80fd\u7f75\u8001\u95c6\uff0c\u56e0\u70ba\u5275\u696d\u516c\u53f8\u6539\u786c\u9ad4\u662f\u7981\u5fcc\uff0c\u53ea\u80fd\u60f3\u8fa6\u6cd5\u628a\u721b\u6a5f\u6551\u6d3b\u3002<\/p>\n\n\n\n\n\n\n\n

    \n\n\n\n

    \u61c9\u7528\u5834\u666f<\/h2>\n\n\n\n

    \u4efb\u52d9\u9700\u6c42\u4e0d\u662f\u7c21\u55ae\u7684\u958b\u500b\u93e1\u982d\u800c\u5df2\uff0c\u800c\u662f\uff1a<\/p>\n\n\n\n

    \n

    \u81c9\u90e8\u8fa8\u8b58\u8a2d\u5099\uff0c\u901a\u904e BCTC \u589e\u5f37\u7d1a\u8a8d\u8b49\uff08\u91d1\u878d\u7b49\u7d1a\uff09<\/strong><\/p>\n<\/blockquote>\n\n\n\n

    \u8981\u6c42\u5305\u542b\uff1a<\/p>\n\n\n\n

      \n
    • \u8fa8\u8b58\u6642\u9593 < 1 \u79d2<\/strong><\/li>\n\n\n\n
    • \u8fa8\u8b58\u7cbe\u5ea6 99.99%<\/strong><\/li>\n\n\n\n
    • \u6d3b\u9ad4\u5b89\u5168\u8a8d\u8b49\uff08\u589e\u5f37\u7d1a\uff09<\/strong><\/li>\n<\/ul>\n\n\n\n
      \n\n\n\n

      \u7cfb\u7d71\u74b0\u5883\u8207\u521d\u6b65\u8a55\u4f30<\/h2>\n\n\n\n

      \u4f5c\u696d\u7cfb\u7d71\uff1aAndroid 10<\/h3>\n\n\n\n

      \u96d6\u7136 QM215 \u5b98\u65b9\u53ea\u63d0\u4f9b Android 9 \u7684 BSP\uff0c\u4f46\u6211\u5011\u9078\u64c7\u5347\u7d1a\u5230 Android 10\uff0c\u76ee\u7684\u662f\u5e0c\u671b\u6e1b\u5c11\u975e\u9810\u671f\u932f\u8aa4\u3002\u4ee3\u50f9\u662f\uff1aAndroid 9~10 \u7684\u7cfb\u7d71\u670d\u52d9\u9810\u8a2d\u5c31\u975e\u5e38\u80a5\uff0c\u5149\u662f\u958b\u6a5f\u5c31\u5403\u6389\u4e86 1.3GB RAM<\/strong>\u3002<\/p>\n\n\n\n

      \u93e1\u982d\u9700\u6c42\uff1aRGB + TOF \u96d9\u93e1\u982d\u4e32\u6d41<\/h3>\n\n\n\n
        \n
      • TOF\uff08Time of Flight\uff09\u662f\u6d3b\u9ad4\u8fa8\u8b58\u7684\u95dc\u9375<\/strong><\/li>\n\n\n\n
      • \u6211\u5011\u63a1\u7528\u7684\u662f IR \u7d50\u69cb\u5149\u5be6\u4f5c<\/strong><\/li>\n\n\n\n
      • \u96d9\u93e1\u982d\u4e32\u6d41 + Frame buffer \u8655\u7406\uff0c\u7d93\u6e2c\u8a66\u7d04\u4f54 950MB \u8a18\u61b6\u9ad4<\/strong><\/li>\n<\/ul>\n\n\n\n

        AI Model \u63a8\u8ad6\u67b6\u69cb<\/h3>\n\n\n\n

        \u5b8c\u6574\u81c9\u90e8\u8fa8\u8b58\u6d41\u7a0b\u5305\u542b\uff1a<\/p>\n\n\n\n

        Face Detect \u2192 Face Align \u2192 Liveness Detection \u2192 Image Quality Check\n\u2192 Feature Extraction \u2192 Feature Matching\n<\/code><\/pre>\n\n\n\n
        \u5176\u4e2d Image Quality Check \u53c8\u7d30\u5206\u70ba\uff1a<\/p>\n\n\n\n
          \n
        • head_pose_model<\/li>\n\n\n\n
        • expression_model<\/li>\n\n\n\n
        • blurriness_model<\/li>\n\n\n\n
        • occlusion_model<\/li>\n\n\n\n
        • cv_over_exposure<\/li>\n<\/ul>\n\n\n\n
          \u9019\u4e9b\u6a21\u7d44\u5927\u591a\u7528 TensorFlow Lite \/ NCNN \u5be6\u4f5c\uff0c\u4e5f\u6709\u4e9b\u7528\u5f71\u50cf\u8655\u7406\u7c21\u5316\u6389\uff0c\u4f46\u4ecd\u7136\u8981\u5403\u6389 \u7d04 300MB \u8a18\u61b6\u9ad4<\/strong>\u3002<\/p>\n\n\n\n
          \n\n\n\n
          \u554f\u984c\u99ac\u4e0a\u6d6e\u73fe<\/h2>\n\n\n\n
          \n
          \u93e1\u982d 950 MB + AI \u6a21\u578b 300 MB + System 1300 MB = 3180 MB > 2000 MB<\/strong><\/p>\n<\/blockquote>\n\n\n\n
          RAM \u76f4\u63a5\u70b8\u88c2\u3002<\/p>\n\n\n\n
          \n
          CPU \u5e7e\u4e4e\u9577\u671f 90% \u6eff\u8f09<\/strong>\uff08A53 x4 \u6838\u5fc3\uff09<\/p>\n<\/blockquote>\n\n\n\n
          AI \u63a8\u8ad6 + \u5f71\u50cf\u8655\u7406\u58d3\u529b\u6975\u5927\u3002<\/p>\n\n\n\n
          \n
          \u76ee\u524d\u63a8\u8ad6 + \u93e1\u982d\u7e3d\u8655\u7406\u6642\u9593 = 1.5 \u79d2<\/strong><\/p>\n<\/blockquote>\n\n\n\n
          \u8ddd\u96e2 <1 \u79d2\u6709\u9ede\u8ddd\u96e2\u3002<\/p>\n\n\n\n
          \n\n\n\n
          \u6211\u5011\u7684\u65b9\u5411<\/h2>\n\n\n\n
            \n
          • \u8a18\u61b6\u9ad4\u7bc0\u7701\u662f\u7b2c\u4e00\u8981\u52d9<\/strong><\/li>\n\n\n\n
          • \u904b\u7b97\u901f\u5ea6\u512a\u5316\u5fc5\u9808\u540c\u6b65\u9032\u884c<\/strong><\/li>\n<\/ul>\n\n\n\n
            \u56e0\u70ba\u6211\u5011\u6709\u4fee\u6539 AOSP \u7684\u6b0a\u9650\uff0c\u7b2c\u4e00\u6b65\u53ef\u4ee5\u5f9e\u780d\u6389\u6c92\u7528\u7684 system service \u958b\u59cb\u3002<\/p>\n\n\n\n
            \u4f46\u5728\u90a3\u4e4b\u524d\uff0c\u8acb\u8a18\u5f97\uff1a<\/p>\n\n\n\n
            \n
            \u98db\u6a5f\u8d77\u98db\u524d\uff0c\u5148\u88dd\u597d\u5100\u8868\u677f\u3002<\/p>\n<\/blockquote>\n\n\n\n
            \u6c92\u6709\u8a18\u61b6\u9ad4\u4f7f\u7528\u76e3\u63a7\u3001\u6c92\u6709\u6548\u80fd\u76e3\u63a7\u5de5\u5177\uff0c\u4e00\u5207\u90fd\u662f\u778e\u5fd9\u3002<\/p>\n\n\n\n
            <\/h2>\n\n\n\n\n\n\n\n
            A215\u662f\u4e00\u6b3e\u9ad8\u901a\u65bc 2019\u5e74\u767c\u552e\u7684\u4f4e\u6548\u80fdCPU\uff0c\u4e3b\u8981\u91dd\u5c0d\u4f4e\u50f9\u667a\u6167\u578b\u624b\u6a5f\u548c\u7269\u806f\u7db2\u8a2d\u5099\uff0c\u512a\u9ede\u662f\u529f\u8017\u8f03\u4f4e\uff0c\u652f\u63f4\u96d9\u93e1\u982d\uff0c\u70ba 64bit \u67b6\u69cb\uff0c\u4f46\u7f3a\u9ede\u662f\u6700\u591a\u53ea\u652f\u63f4 \uff13GB ram\uff0c\u751a\u81f3\u70ba\u4e86\u7701\u9322\uff0c\u8001\u95c6\u53ea\u9858\u610f\u88dd2GB\uff0c\u6b64\u5916 A215 \u6c92\u6709NPU \u4f46\u5177\u6709 GPU and CPU(1.3G)<\/p>\n\n\n<\/p>\n
            \n\n\n
            \u4ee5\u4e0a\u662f\u4f60\u8eab\u7232\u6280\u8853\u8ca0\u8cac\u4eba\u65bc\uff12\uff10\uff12\uff14\u5e74\u62ff\u5230\u7684\u4e00\u500b\u88dd\u7f6e\uff0c\u8981\u60f3\u60f3\u9019\u662f\u4e00\u500b4\u5e74\u524d\u767c\u552e\u7684\u5165\u9580\u7d1a\u8655\u7406\u5668\uff0c\u4f46\u662f\u8eab\u70ba\u5c08\u696d\u7684\u4f60\u4e0d\u80fd\u99ac\u4e0a\u958b\u59cb\u627e\u8001\u95c6\u7b97\u5e33\uff0c\u7562\u7adf\u516c\u53f8\u627e\u4f60\u4f86\u662f\u89e3\u6c7a\u554f\u984c\uff0c\u800c\u5728\u65b0\u5275\u516c\u53f8\uff0c\u8981\u82b1\u9322\u6539\u786c\u9ad4\u901a\u5e38\u662f\u8001\u95c6\u6700\u4e0d\u9858\u610f\u505a\u7684\u4e8b\u60c5\u3002\u91cd\u9ede\u662f\u4e0d\u7ba1\u662f\u5565\u721b\u88dd\u7f6e\uff0c\u53ea\u8981\u61c9\u7528\u7684\u60c5\u6cc1\u9069\u7576\uff0c\u5c31\u662f\u53ef\u4ee5\u7528<\/p>\n\n\n<\/p>\n
            \n\n\n
            \u61c9\u7528\u5834\u666f-\u4eba\u81c9\u8fa8\u8b58\u88dd\u7f6e\u4e14\u80fd\u901a\u904e BCTC \u589e\u5f37\u7d1a\u8a8d\u8b49\uff08\u8fa8\u8b58\u6642\u9593<1\u79d2\uff0c\u8fa8\u8b58\u7cbe\u5ea6\u9ad8\u905499.99%\uff0c\u6d3b\u9ad4\u5b89\u5168\u8a8d\u8b49\uff08\u589e\u5f37\u7d1a\uff09\uff0c\u9054\u5230\u91d1\u878d\u7d1a\u652f\u4ed8\u6a19\u6e96\uff09<\/strong><\/p>\n\n\n<\/p>\n
            \n\n\n
            Android 10 <\/strong>– \u4e3b\u8981\u662f\u56e0\u70ba QM215 \u5b98\u65b9\u53ea\u91dd\u5c0d android 9 \u63d0\u4f9bBSP\uff0c\u70ba\u4e86\u964d\u4f4e\u53ef\u80fd\u51fa\u73fe\u7684\u975e\u9810\u671f\u60c5\u6cc1\uff0c\u7136\u800c\u4ee3\u50f9\u5c31\u662f android 9 system service \u8dd1\u8d77\u4f86\u5c31\u6703\u6d88\u8017 1.3 GB memory used<\/p>\n\n\n<\/p>\n
            \n\n\n
            \u96d9\u93e1\u982d RGB and TOF Streaming – \u4eba\u81c9\u8fa8\u8b58\u53ea\u9760\u55ae\u500b RGB \u93e1\u982d\u662f\u7121\u6cd5\u9054\u5230\u6d3b\u9ad4\u8fa8\u8b58 Liveness Check\uff0c\u6240\u4ee5\u5fc5\u9808\u5229\u7528TOF\uff08Time of Flight\uff09\uff0c\u9019\u90e8\u5206\u6709\u597d\u5e7e\u7a2e\u5be6\u4f5c\u65b9\u5f0f\uff0c\u4f46\u6211\u5011\u662f\u63a1\u7528IR\uff0c\u9019\u90e8\u5206\u6211\u5011\u7b49\u540c\u65bc\u9700\u8981\u540c\u6642\u8655\u7406\u5169\u500b\u93e1\u982d\u7684\u8cc7\u6599\u6d41\uff0c\u9019\u610f\u5473\u8005 cpu \u904b\u7b97\u8cc7\u6e90\u8ddf memory \u90fd\u6703\u9ad8\u901f\u6d88\u8017\uff0e\u9019\u90e8\u5206\u7d93\u904e\u6e2c\u8a66\u7d04\u4f7f\u7528950mb<\/p>\n\n\n<\/p>\n
            \n\n\n
            \u4eba\u81c9\u8fa8\u8b58 AI Model – \u5728\u5927\u90e8\u5206\u7684\u60c5\u6cc1\u4e0b\uff0c\u5c31\u7b97\u91dd\u5c0d EDGE device \u53bb\u512a\u5316\u6a21\u578b\uff0c\u4e5f\u53ea\u662f\u76f8\u5c0d\u5927\u578bserver\u4e0a\u7684\u5dee\u7570\uff0c\u6c92\u8fa6\u6cd5\u771f\u7684\u58d3\u5230\u5f88\u4f4e\u7684\u904b\u7b97\u8cc7\u6e90\u53bb\u8dd1\uff0c\u7d93\u904e\u591a\u65b9\u8a55\u4f30\u63a1\u7528 tensorflow lite, ncnn \uff08\u60f3\u7576\u5e74\u9084\u7528 caffe2\uff09\uff0c\u52a0\u4e0a\u4e00\u500b\u5b8c\u6574\u7684\u4eba\u81c9\u8b58\u5225\u6d41\u7a0b\uff0c\u4e26\u975e\u4e00\u500b\u55ae\u4e00\u6a21\u578b\u5c31\u53ef\u4ee5\u5b8c\u6210\uff0c\u4ee5\u4e0b\u662f\u6700\u57fa\u672c\u7684\u6d41\u7a0b<\/p>\n\n\n<\/p>\n
            \n\n\n
            Face Detect \u2192 Face Align \u2192 Liveness Detection \u2192 Image Quality Check\uff08head_pose_model<\/strong>\uff0cexpression_model\uff0cblurriness_model\uff0cocclusion_model\uff0ccv_over_exposure\uff09 \u2192 Feature Extraction \u2192 Feature Matching<\/strong>.<\/p>\n\n\n<\/p>\n
            \n\n\n
            \u53ef\u80fd\u4e26\u975e\u6bcf\u500b\u90fd\u9700\u8981ai infer\uff0c\u6709\u4e9b\u4f7f\u7528\u50b3\u7d71\u5f71\u50cf\u8655\u7406\u65b9\u59cb\u5c31\u5f88\u5920\u4e86\uff0c\u4f46\u7d9c\u5408\u8a55\u4f30\u4e0b\u4f86\u7d04\u5316\u4f7f\u7528 300 MB \u7684\u8a18\u61b6\u9ad4\uff0c\u4ee5\u53ca\u5927\u91cf\u7684 CPU \u904b\u7b97\u8cc7\u6e90<\/p>\n\n\n\n
            \u80cc\u666f\u4ea4\u4ee3\u5b8c\u7562\u4e86\uff0c\u63a5\u4e0b\u6211\u5011\u99ac\u4e0a\u5c31\u767c\u73fe<\/p>\n\n\n\n
            memory\u7206\u70b8 -> \u93e1\u982d 950 mb + ai model 300 mb + system 1300 mb > 2000 mb RAM<\/p>\n\n\n\n
            CPU \u8ca0\u8377\u5e7e\u4e4e\u96a8\u6642\u90fd\u5728 90%\uff0c\u4e3b\u8981\u4f86\u81ea\u65bc AI \u63a8\u7406\u8207\u5f71\u50cf\u8655\u7406\uff084x Cortex-A53 \u5168\u6838\u5fc3\u5e7e\u4e4e\u90fd\u6eff\u8f09\uff09\u3002\uff0c\u76ee\u524d\u6574\u500b\u4eba\u81c9\u8fa8\u8b58\u6d41\u7a0b\uff08\u5305\u542b\u5f71\u50cf\u8655\u7406 + AI \u63a8\u7406\uff09\u7d04\u9700 1.5s\u3002<\/p>\n\n\n\n
            \u7bc0\u7701\u8a18\u61b6\u9ad4\uff0b\u512a\u5316\u8a08\u7b97\u901f\u5ea6\u8b8a\u6210\u6211\u5011\u7684\u7b2c\u4e00\u9805\u5de5\u4f5c\uff0c\u91dd\u5c0d\u8a18\u61b6\u9ad4\u53ef\u4ee5\u5148\u5f9e\u7cfb\u7d71\u4e0b\u624b\uff0c\u7562\u7adf\u6211\u5011\u64c1\u6709\u6574\u500b AOSP \u7684\u4fee\u6539\u6b0a\u5229\uff0c\u4f46\u5728\u9019\u4e4b\u524d\u53ef\u80fd\u5927\u5bb6\u9700\u8981\u5148\u77e5\u9053\u600e\u9ebc\u8a55\u4f30\u8a18\u61b6\u9ad4\u4f7f\u7528\uff0c\u7562\u7adf\u6c92\u6709\u5100\u8868\u677f\u7684\u98db\u6a5f\uff0c\u6c92\u4eba\u6562\u958b\uff0c\u6c92\u6709\u8a55\u4f30\u6a5f\u5236\uff0c\u600e\u77e5\u9053\u6210\u6548\u597d\u58de<\/p>\n\n\n\n
            <\/code><\/pre>\n\n\n<\/p>\n
            \n\n\n
            \n\n\n\n
            <\/h3>\n","protected":false},"excerpt":{"rendered":"
            \u721b\u88dd\u7f6e\u662f\u5565\uff1f<\/p>\n
            Qualcomm Sanpdragon QM215\u00a0 armv8a \uff0b 2GB Memory<\/p>\n
            A215\u662f\u4e00\u6b3e\u9ad8\u901a\u65bc 2019\u5e74\u767c\u552e\u7684\u4f4e\u6548\u80fdCPU\uff0c\u4e3b\u8981\u91dd\u5c0d\u4f4e\u50f9\u667a\u6167\u578b\u624b\u6a5f\u548c\u7269\u806f\u7db2\u8a2d\u5099\uff0c\u512a\u9ede\u662f\u529f\u8017\u8f03\u4f4e\uff0c\u652f\u63f4\u96d9\u93e1\u982d\uff0c\u70ba 64bit \u67b6\u69cb\uff0c\u4f46\u7f3a\u9ede\u662f\u6700\u591a\u53ea\u652f\u63f4 \uff13GB ram\uff0c\u751a\u81f3\u70ba\u4e86\u7701\u9322\uff0c\u8001\u95c6\u53ea\u9858\u610f\u88dd2GB\uff0c\u6b64\u5916 A215 \u6c92\u6709NPU \u4f46\u5177\u6709 GPU and CPU(1.3G)<\/p>\n","protected":false},"author":1,"featured_media":567,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[80],"tags":[],"_links":{"self":[{"href":"https:\/\/boochlin.com\/index.php?rest_route=\/wp\/v2\/posts\/401"}],"collection":[{"href":"https:\/\/boochlin.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/boochlin.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/boochlin.com\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/boochlin.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=401"}],"version-history":[{"count":25,"href":"https:\/\/boochlin.com\/index.php?rest_route=\/wp\/v2\/posts\/401\/revisions"}],"predecessor-version":[{"id":588,"href":"https:\/\/boochlin.com\/index.php?rest_route=\/wp\/v2\/posts\/401\/revisions\/588"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/boochlin.com\/index.php?rest_route=\/wp\/v2\/media\/567"}],"wp:attachment":[{"href":"https:\/\/boochlin.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=401"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/boochlin.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=401"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/boochlin.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=401"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}