{"id":64,"date":"2026-05-06T12:08:01","date_gmt":"2026-05-06T12:08:01","guid":{"rendered":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/?page_id=64"},"modified":"2026-05-06T14:03:19","modified_gmt":"2026-05-06T14:03:19","slug":"experiments","status":"publish","type":"page","link":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/project2\/experiments\/","title":{"rendered":"Experiments"},"content":{"rendered":"<section>\n  <h2>Cooperative Foundation Models for Object Detection Active Learning<\/h2>\n\n  <nav>\n    <a href=\"\/2026teamf11\/project2\/\">Introduction<\/a> |\n    <a href=\"\/2026teamf11\/project2\/related-work\/\">Related Work<\/a> |\n    <a href=\"\/2026teamf11\/project2\/method\/\">Method<\/a> |\n    <strong>Experiments<\/strong> |\n    <a href=\"\/2026teamf11\/project2\/resources\/\">Resources<\/a> |\n    <a href=\"\/2026teamf11\/project2\/team\/\">Team<\/a>\n  <\/nav>\n\n  <hr>\n\n  <section>\n    <h2>Experiments<\/h2>\n\n    <h3>Experimental Setup<\/h3>\n\n    <p>\n      We evaluate CoMODAL on two object detection benchmarks: <strong>DIOR<\/strong> for remote-sensing\n      images and <strong>COCO<\/strong> for general object detection. We focus on one frequent target class\n      in each dataset: <strong>vehicle<\/strong> for DIOR and <strong>person<\/strong> for COCO.\n    <\/p>\n\n    <p>\n      In the current experiments, we use <strong>RetinaNet R50<\/strong> as the target object detector.\n      CoMODAL is compared with common active learning baselines, including Random Sampling,\n      Entropy-based Sampling, Coreset, PPAL, and DivProto.\n    <\/p>\n\n    <h3>Results<\/h3>\n\n    <p>\n      CoMODAL consistently outperforms the active learning baselines on both DIOR-vehicle and\n      COCO-person under the same annotation budget. This shows that cooperative foundation models\n      can improve sample selection for RetinaNet-based object detection.\n    <\/p>\n\n    <p>\n      On average, CoMODAL improves performance by <strong>3.1%<\/strong> on DIOR-vehicle with RetinaNet\n      and <strong>4.3%<\/strong> on COCO-person with RetinaNet.\n    <\/p>\n\n<\/section>\n<\/section>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"292\" src=\"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-content\/uploads\/sites\/153\/2026\/05\/results-1024x292.jpg\" alt=\"\" class=\"wp-image-154\" style=\"width:800px\" srcset=\"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-content\/uploads\/sites\/153\/2026\/05\/results-1024x292.jpg 1024w, https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-content\/uploads\/sites\/153\/2026\/05\/results-300x85.jpg 300w, https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-content\/uploads\/sites\/153\/2026\/05\/results-768x219.jpg 768w, https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-content\/uploads\/sites\/153\/2026\/05\/results-1536x438.jpg 1536w, https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-content\/uploads\/sites\/153\/2026\/05\/results.jpg 2022w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n<section>\n<section>\n\n    <h3>Automatic Annotation Efficiency<\/h3>\n\n    <p>\n      We also evaluate the quality of bounding boxes generated by the foundation-model annotation\n      committee. For RetinaNet experiments, the generated boxes match a large portion of the ground\n      truth boxes, with average matching rates of <strong>87.14%<\/strong> on DIOR-vehicle and\n      <strong>87.65%<\/strong> on COCO-person.\n    <\/p>\n\n    <p>\n      Since humans only need to approve or reject proposed boxes, the annotation process becomes much\n      faster than drawing boxes manually. Based on the estimated time difference between box drawing\n      and binary verification, CoMODAL can substantially reduce query annotation time.\n    <\/p>\n\n<\/section>\n<\/section>\n\n\n<figure class=\"wp-block-table is-style-stripes has-medium-font-size\"><table class=\"has-contrast-color has-base-background-color has-text-color has-background has-link-color has-fixed-layout\"><thead><tr><th>Statistics Name<\/th><th class=\"has-text-align-center\" data-align=\"center\">COCO (Person)<\/th><th class=\"has-text-align-center\" data-align=\"center\">DIOR (Vehicle)<\/th><\/tr><\/thead><tbody><tr><td>GTs average matching rate<\/td><td class=\"has-text-align-center\" data-align=\"center\">87.65%<\/td><td class=\"has-text-align-center\" data-align=\"center\">87.14%<\/td><\/tr><tr><td>FMs annotations approval rate<\/td><td class=\"has-text-align-center\" data-align=\"center\">31.08%<\/td><td class=\"has-text-align-center\" data-align=\"center\">26.41%<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n<section>\n<section>\n    <h3>Summary<\/h3>\n\n    <p>\n      The RetinaNet results show that CoMODAL improves active learning performance while reducing\n      annotation effort. By combining foundation-model assistance with binary human feedback, the\n      framework provides a more efficient way to train object detectors under limited annotation budgets.\n    <\/p>\n  <\/section>\n<\/section>\n\n\n<div style=\"height:100px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n","protected":false},"excerpt":{"rendered":"<p>Cooperative Foundation Models for Object Detection Active Learning Introduction | Related Work | Method | Experiments | Resources | Team Experiments Experimental Setup We evaluate CoMODAL on two object detection benchmarks: DIOR for remote-sensing images and COCO for general object detection. We focus on one frequent target class in each dataset: vehicle for DIOR and [&hellip;]<\/p>\n","protected":false},"author":288,"featured_media":0,"parent":34,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-64","page","type-page","status-publish","hentry"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.4 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>Experiments - Disentangled Gaussian Transformers for Animatable Humans | Cooperative Foundation Models For Object Detection Active Learning<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/project2\/experiments\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Experiments - Disentangled Gaussian Transformers for Animatable Humans | Cooperative Foundation Models For Object Detection Active Learning\" \/>\n<meta property=\"og:description\" content=\"Cooperative Foundation Models for Object Detection Active Learning Introduction | Related Work | Method | Experiments | Resources | Team Experiments Experimental Setup We evaluate CoMODAL on two object detection benchmarks: DIOR for remote-sensing images and COCO for general object detection. We focus on one frequent target class in each dataset: vehicle for DIOR and [&hellip;]\" \/>\n<meta property=\"og:url\" content=\"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/project2\/experiments\/\" \/>\n<meta property=\"og:site_name\" content=\"Disentangled Gaussian Transformers for Animatable Humans | Cooperative Foundation Models For Object Detection Active Learning\" \/>\n<meta property=\"article:modified_time\" content=\"2026-05-06T14:03:19+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-content\/uploads\/sites\/153\/2026\/05\/results.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"2022\" \/>\n\t<meta property=\"og:image:height\" content=\"576\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data1\" content=\"1 minute\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/project2\\\/experiments\\\/\",\"url\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/project2\\\/experiments\\\/\",\"name\":\"Experiments - Disentangled Gaussian Transformers for Animatable Humans | Cooperative Foundation Models For Object Detection Active Learning\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/project2\\\/experiments\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/project2\\\/experiments\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/wp-content\\\/uploads\\\/sites\\\/153\\\/2026\\\/05\\\/results-1024x292.jpg\",\"datePublished\":\"2026-05-06T12:08:01+00:00\",\"dateModified\":\"2026-05-06T14:03:19+00:00\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/project2\\\/experiments\\\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/project2\\\/experiments\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/project2\\\/experiments\\\/#primaryimage\",\"url\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/wp-content\\\/uploads\\\/sites\\\/153\\\/2026\\\/05\\\/results.jpg\",\"contentUrl\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/wp-content\\\/uploads\\\/sites\\\/153\\\/2026\\\/05\\\/results.jpg\",\"width\":2022,\"height\":576},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/project2\\\/experiments\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Project 2\",\"item\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/project2\\\/\"},{\"@type\":\"ListItem\",\"position\":3,\"name\":\"Experiments\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/#website\",\"url\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/\",\"name\":\"Disentangled Gaussian Transformers for Animatable Humans | Cooperative Foundation Models For Object Detection Active Learning\",\"description\":\"\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf11\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Experiments - Disentangled Gaussian Transformers for Animatable Humans | Cooperative Foundation Models For Object Detection Active Learning","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/project2\/experiments\/","og_locale":"en_US","og_type":"article","og_title":"Experiments - Disentangled Gaussian Transformers for Animatable Humans | Cooperative Foundation Models For Object Detection Active Learning","og_description":"Cooperative Foundation Models for Object Detection Active Learning Introduction | Related Work | Method | Experiments | Resources | Team Experiments Experimental Setup We evaluate CoMODAL on two object detection benchmarks: DIOR for remote-sensing images and COCO for general object detection. We focus on one frequent target class in each dataset: vehicle for DIOR and [&hellip;]","og_url":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/project2\/experiments\/","og_site_name":"Disentangled Gaussian Transformers for Animatable Humans | Cooperative Foundation Models For Object Detection Active Learning","article_modified_time":"2026-05-06T14:03:19+00:00","og_image":[{"width":2022,"height":576,"url":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-content\/uploads\/sites\/153\/2026\/05\/results.jpg","type":"image\/jpeg"}],"twitter_card":"summary_large_image","twitter_misc":{"Est. reading time":"1 minute"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/project2\/experiments\/","url":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/project2\/experiments\/","name":"Experiments - Disentangled Gaussian Transformers for Animatable Humans | Cooperative Foundation Models For Object Detection Active Learning","isPartOf":{"@id":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/#website"},"primaryImageOfPage":{"@id":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/project2\/experiments\/#primaryimage"},"image":{"@id":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/project2\/experiments\/#primaryimage"},"thumbnailUrl":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-content\/uploads\/sites\/153\/2026\/05\/results-1024x292.jpg","datePublished":"2026-05-06T12:08:01+00:00","dateModified":"2026-05-06T14:03:19+00:00","breadcrumb":{"@id":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/project2\/experiments\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/project2\/experiments\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/project2\/experiments\/#primaryimage","url":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-content\/uploads\/sites\/153\/2026\/05\/results.jpg","contentUrl":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-content\/uploads\/sites\/153\/2026\/05\/results.jpg","width":2022,"height":576},{"@type":"BreadcrumbList","@id":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/project2\/experiments\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/"},{"@type":"ListItem","position":2,"name":"Project 2","item":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/project2\/"},{"@type":"ListItem","position":3,"name":"Experiments"}]},{"@type":"WebSite","@id":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/#website","url":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/","name":"Disentangled Gaussian Transformers for Animatable Humans | Cooperative Foundation Models For Object Detection Active Learning","description":"","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"}]}},"_links":{"self":[{"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-json\/wp\/v2\/pages\/64","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-json\/wp\/v2\/users\/288"}],"replies":[{"embeddable":true,"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-json\/wp\/v2\/comments?post=64"}],"version-history":[{"count":6,"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-json\/wp\/v2\/pages\/64\/revisions"}],"predecessor-version":[{"id":158,"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-json\/wp\/v2\/pages\/64\/revisions\/158"}],"up":[{"embeddable":true,"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-json\/wp\/v2\/pages\/34"}],"wp:attachment":[{"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf11\/wp-json\/wp\/v2\/media?parent=64"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}