{"id":9983,"date":"2026-07-08T16:40:09","date_gmt":"2026-07-08T08:40:09","guid":{"rendered":"https:\/\/infernews.com\/blog\/self-supervised-learning-for-spatial-perception\/"},"modified":"2026-07-08T16:40:09","modified_gmt":"2026-07-08T08:40:09","slug":"self-supervised-learning-for-spatial-perception","status":"publish","type":"post","link":"https:\/\/infernews.com\/blog\/self-supervised-learning-for-spatial-perception\/","title":{"rendered":"LingBot-Vision \u88dc\u5f37\u5bc6\u96c6\u7a7a\u9593\u611f\u77e5"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/07\/pasted-5a6bc74ba412.jpg\" alt=\"Boundary-centric masked modeling\"><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">\u505a\u6df1\u5ea6\u4f30\u8a08\u3001\u8a9e\u610f\u5206\u5272\u6216\u8005\u5f71\u7247\u7269\u4ef6\u5206\u5272\u6642\uff0c\u6700\u9ebb\u7169\u5f80\u5f80\u5514\u4fc2\u6709\u5187\u5927\u6a21\u578b\uff0c\u800c\u662f\u7de8\u78bc\u5668\u62bd\u51fa\u7684\u7279\u5fb5\u5920\u5514\u5920\u8cbc\u8fd1\u7269\u4ef6\u8f2a\u5ed3\u3002LingBot-Vision\u5c6c\u65bc<strong>\u6a21\u578b<\/strong>\uff0c\u66f4\u6e96\u78ba\u5730\u8aaa\u662f\u81ea\u76e3\u7763\u9810\u8a13\u7df4\u7684 Vision Transformer \u8996\u89ba\u9aa8\u5e79\uff0c\u8655\u7406\u7684\u662f\u5bc6\u96c6\u7a7a\u9593\u611f\u77e5\u88cf\u9762\u300c\u8a9e\u610f\u6709\u4e86\uff0c\u4f46\u908a\u754c\u5514\u5920\u6e96\u300d\u9019\u500b\u8001\u554f\u984c\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u5b83\u7684\u53d6\u5411\u5e7e\u660e\u78ba\uff1a\u5514\u4fc2\u4e00\u5473\u8ffd\u6c42\u5206\u985e\u5f0f\u8a9e\u610f\u8868\u793a\uff0c\u800c\u662f\u7528 <strong>masked boundary modeling<\/strong> \u53bb\u903c\u6a21\u578b\u540c\u6642\u4fdd\u7559\u7a7a\u9593\u7d50\u69cb\u8207\u8a9e\u610f\u8a0a\u606f\u3002\u51cd\u7d50\u5f8c\u7684 patch tokens \u5df2\u7d93\u53ef\u4ee5\u76f4\u63a5\u652f\u63f4\u8f15\u91cf readout\uff0c\u6db5\u84cb depth estimation\u3001semantic segmentation\u3001video object segmentation\uff0c\u4ea6\u4f5c\u70ba LingBot-Depth 2.0 \u7684 visual encoder \u521d\u59cb\u5316\uff0c\u9019\u7a2e\u8a2d\u8a08\u6bd4\u8d77\u53ea\u504f\u91cd\u5168\u5c40\u8a9e\u610f\u7684 ViT \u9aa8\u5e79\uff0c\u66f4\u9069\u5408\u9700\u8981\u9010\u50cf\u7d20\u5224\u65b7\u7684\u5de5\u4f5c\u6d41\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u9019\u500b\u9805\u76ee\u8f03\u63a5\u8fd1\u300c\u53d6\u7528\u9810\u8a13\u7df4\u9aa8\u5e79\u518d\u63a5\u4e0b\u6e38\u4efb\u52d9\u300d\u7684\u7528\u6cd5\uff0c\u800c\u5514\u4fc2\u5373\u88dd\u5373\u7528\u7684\u5b8c\u6574\u61c9\u7528\u3002\u6a21\u578b\u5df2\u653e\u5230 Hugging Face \u8207 ModelScope\uff0c\u8f03\u5408\u7406\u7684\u7406\u89e3\u65b9\u5f0f\uff0c\u662f\u628a\u4e0d\u540c\u5c3a\u5bf8\u7684 LingBot-Vision \u6b0a\u91cd\u63a5\u5165\u73fe\u6709 dense prediction pipeline\uff0c\u5148\u6e2c frozen features \u7684\u8868\u73fe\uff0c\u518d\u6c7a\u5b9a\u9700\u5514\u9700\u8981\u984d\u5916\u5fae\u8abf\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u91cd\u9ede\u4e0d\u5728\u751f\u6210\u5167\u5bb9\uff0c\u800c\u5728\u63d0\u9ad8 dense spatial perception \u7684\u7279\u5fb5\u54c1\u8cea<\/li>\n<li>\u5df2\u516c\u958b\u591a\u500b\u76f8\u95dc\u6a21\u578b\uff1aViT-S\/16\u3001ViT-Base\u3001ViT-L\/16\uff0c\u4ee5\u81f3 1.1B \u53c3\u6578\u7684 ViT-g\/16<\/li>\n<li>\u652f\u63f4\u7684\u65b9\u5411\u5305\u62ec depth estimation\u3001semantic segmentation\u3001video object segmentation\u3001depth completion<\/li>\n<li>\u8207\u540c\u985e\u505a\u6cd5\u76f8\u6bd4\uff0c\u66f4\u91cd\u8996 boundary-faithful features\uff0c\u800c\u5514\u4fc2\u53ea\u5f37\u5316\u9ad8\u5c64\u8a9e\u610f\u8868\u793a<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">\u53d7\u76ca\u6700\u5927\u7684\u6703\u662f\u505a\u6a5f\u68b0\u4eba\u8996\u89ba\u30013D \u611f\u77e5\u3001\u5f71\u50cf\u7406\u89e3\u57fa\u5efa\u7684\u5718\u968a\uff0c\u5c24\u5176\u7576\u4f60\u624b\u4e0a\u5df2\u6709 segmentation \u6216 depth \u9805\u76ee\uff0c\u53ea\u5dee\u4e00\u500b\u66f4\u7a69\u5b9a\u7684 encoder\u3002\u6027\u80fd\u65b9\u9762\uff0cREADME \u7528\u300csubstantial performance gains\u300d\u5f62\u5bb9 LingBot-Depth 2.0 \u5728\u63db\u4e0a LingBot-Vision \u7de8\u78bc\u5668\u5f8c\u7684\u63d0\u5347\uff0c\u4f46\u516c\u958b\u5167\u5bb9\u672a\u5217\u51fa\u5b8c\u6574\u57fa\u6e96\u6578\u5b57\uff0c\u6240\u4ee5\u73fe\u968e\u6bb5\u8f03\u503c\u5f97\u5148\u7559\u610f\u5176\u7279\u5fb5\u53ef\u9077\u79fb\u6027\uff0c\u4ee5\u53ca\u5728\u908a\u754c\u654f\u611f\u4efb\u52d9\u4e0a\u7684\u6f5b\u529b\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/technology.robbyant.com\/lingbot-vision\" rel=\"noopener noreferrer\"><strong>\u9805\u76ee\u4e3b\u9801<\/strong><\/a> \u00b7 <a href=\"https:\/\/github.com\/Robbyant\/lingbot-vision\" rel=\"noopener noreferrer\"><strong>GitHub<\/strong><\/a> \u00b7 <a href=\"https:\/\/huggingface.co\/collections\/robbyant\/lingbot-vision\" rel=\"noopener noreferrer\"><strong>\u6a21\u578b<\/strong><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u60f3\u505a\u6df1\u5ea6\u4f30\u8a08\u6216\u8a9e\u610f\u5206\u5272\uff0c\u4f46\u53c8\u5514\u60f3\u6bcf\u500b\u4e0b\u6e38\u4efb\u52d9\u90fd\u91cd\u8a13\u8996\u89ba\u7de8\u78bc\u5668\uff0cLingBot-Vision\u6b63\u662f\u91dd\u5c0d\u9019\u500b\u5361\u4f4d\u800c\u4f86\u3002\u5b83\u628a\u908a\u754c\u3001\u5f62\u72c0\u540c\u8a9e\u610f\u5340\u57df\u4e00\u4f75\u5b78\u597d\uff0c\u91cd\u9ede\u653e\u5728\u66f4\u53ef\u9760\u7684\u7a7a\u9593\u7279\u5fb5\u3002<\/p>\n","protected":false},"author":8,"featured_media":9982,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ai_generated_summary":"","footnotes":""},"categories":[133,31,132,119,120,76,184,204],"tags":[],"class_list":["post-9983","post","type-post","status-publish","format-standard","hentry","category-133","category-video","category-3d","category-119","category-120","category-76","category-robotic","category-visionlanguageaction"],"_links":{"self":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/9983","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/users\/8"}],"replies":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/comments?post=9983"}],"version-history":[{"count":0,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/9983\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media\/9982"}],"wp:attachment":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media?parent=9983"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/categories?post=9983"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/tags?post=9983"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}