{"id":66,"date":"2024-05-10T04:55:26","date_gmt":"2024-05-10T04:55:26","guid":{"rendered":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/?page_id=66"},"modified":"2024-12-14T04:08:48","modified_gmt":"2024-12-14T04:08:48","slug":"method","status":"publish","type":"page","link":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/method\/","title":{"rendered":"Method"},"content":{"rendered":"\n

Architecture<\/strong><\/h2>\n\n\n\n

Match Anything<\/strong><\/p>\n\n\n

\n
\"\"<\/figure>\n<\/div>\n\n\n
\n
\n

1. Simplify the architecture to be more generalizable
\u00b7 <\/strong>Include more dataset for training<\/p>\n\n\n\n

2. Only predict flow in the co-visible region<\/p>\n\n\n\n

\u00b7 <\/strong>This makes matching a 2D problem, as the network don\u2019t need to reason about occlusions\/geometry!<\/p>\n<\/div>\n<\/div>\n\n\n\n

\"\"<\/figure>\n\n\n\n
\n

Co-visible Mask Generation<\/strong><\/p>\n\n\n\n

Project pixels from camera 1 into 3D space and then back to camera 2<\/p>\n\n\n\n

FoV mask<\/strong>: If the coordinates of projected pixel is within image boundaries, then it\u2019s in FOV
Occlusion mask<\/strong>: If the depth of projected pixel is close to the real depth of that pixel, then it\u2019s visible<\/p>\n\n\n\n

\"\"<\/figure>\n<\/div>\n\n\n\n

Sampling Method<\/strong><\/h2>\n\n\n\n

To get image pairs, we need to design a sampling method.<\/p>\n\n\n\n

\n

1. Accumulate Pointcloud:<\/strong><\/p>\n\n\n\n

From posed depth image, upproject into point cloud and accumulate.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

2. Voxelize:<\/strong><\/p>\n\n\n\n

Voxel down-sample point clouds & camera positions to create enumerable scene representation.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

3. Calculate Covisibility:<\/strong><\/p>\n\n\n\n

For all camera position to all voxels, determine if the camera can see the voxel. Save to a list.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

4. Generate Samples:<\/strong><\/p>\n\n\n\n

Randomly select a base camera and a target voxel, filter all candidate camera position that can have required angle with the base camera when looking at the target voxel.<\/p>\n\n\n\n

Score all candidates based on visibility from the preprocessed covisibility list. Keep N candidates and add to pair list.<\/p>\n\n\n\n

\"\"<\/figure>\n<\/div>\n\n\n\n
\"\"<\/figure>\n\n\n\n

Utilizing match anything for reconstruction<\/strong><\/h2>\n\n\n\n

To achieve precise mapping and odometry, we build upon state of art method MAC-VO. Essentially we replace the current FlowFormer based frontend with Match Anything. We have started to integrate it and the results look promising.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n
\"\"<\/figure>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n","protected":false},"excerpt":{"rendered":"

Architecture Match Anything 1. Simplify the architecture to be more generalizable \u00b7 Include more dataset for training 2. Only predict flow in the co-visible region \u00b7 This makes matching a 2D problem, as the network don\u2019t need to reason about occlusions\/geometry! Co-visible Mask Generation Project pixels from camera 1 into 3D space and then back … <\/p>\n

Continue reading “Method”<\/span><\/a><\/p>\n","protected":false},"author":221,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-66","page","type-page","status-publish","hentry"],"yoast_head":"\nMethod - Towards Universal State Estimation and Reconstruction in the Wild<\/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\/2024team15\/method\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Method - Towards Universal State Estimation and Reconstruction in the Wild\" \/>\n<meta property=\"og:description\" content=\"Architecture Match Anything 1. Simplify the architecture to be more generalizable \u00b7 Include more dataset for training 2. Only predict flow in the co-visible region \u00b7 This makes matching a 2D problem, as the network don\u2019t need to reason about occlusions\/geometry! Co-visible Mask Generation Project pixels from camera 1 into 3D space and then back … Continue reading "Method"\" \/>\n<meta property=\"og:url\" content=\"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/method\/\" \/>\n<meta property=\"og:site_name\" content=\"Towards Universal State Estimation and Reconstruction in the Wild\" \/>\n<meta property=\"article:modified_time\" content=\"2024-12-14T04:08:48+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/wp-content\/uploads\/sites\/113\/2024\/12\/image.png\" \/>\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=\"4 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2024team15\\\/method\\\/\",\"url\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2024team15\\\/method\\\/\",\"name\":\"Method - Towards Universal State Estimation and Reconstruction in the Wild\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2024team15\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2024team15\\\/method\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2024team15\\\/method\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2024team15\\\/wp-content\\\/uploads\\\/sites\\\/113\\\/2024\\\/12\\\/image.png\",\"datePublished\":\"2024-05-10T04:55:26+00:00\",\"dateModified\":\"2024-12-14T04:08:48+00:00\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2024team15\\\/method\\\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2024team15\\\/method\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2024team15\\\/method\\\/#primaryimage\",\"url\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2024team15\\\/wp-content\\\/uploads\\\/sites\\\/113\\\/2024\\\/12\\\/image.png\",\"contentUrl\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2024team15\\\/wp-content\\\/uploads\\\/sites\\\/113\\\/2024\\\/12\\\/image.png\",\"width\":713,\"height\":766},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2024team15\\\/method\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2024team15\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Method\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2024team15\\\/#website\",\"url\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2024team15\\\/\",\"name\":\"Towards Universal State Estimation and Reconstruction in the Wild\",\"description\":\"Student:Bhuvan Jhamb, Chenwei Lyu | Advisors: Nikhil Keetha, Yuchen Zhang, Dr. Sebastian Scherer\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2024team15\\\/?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":"Method - Towards Universal State Estimation and Reconstruction in the Wild","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\/2024team15\/method\/","og_locale":"en_US","og_type":"article","og_title":"Method - Towards Universal State Estimation and Reconstruction in the Wild","og_description":"Architecture Match Anything 1. Simplify the architecture to be more generalizable \u00b7 Include more dataset for training 2. Only predict flow in the co-visible region \u00b7 This makes matching a 2D problem, as the network don\u2019t need to reason about occlusions\/geometry! Co-visible Mask Generation Project pixels from camera 1 into 3D space and then back … Continue reading \"Method\"","og_url":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/method\/","og_site_name":"Towards Universal State Estimation and Reconstruction in the Wild","article_modified_time":"2024-12-14T04:08:48+00:00","og_image":[{"url":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/wp-content\/uploads\/sites\/113\/2024\/12\/image.png","type":"","width":"","height":""}],"twitter_card":"summary_large_image","twitter_misc":{"Est. reading time":"4 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/method\/","url":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/method\/","name":"Method - Towards Universal State Estimation and Reconstruction in the Wild","isPartOf":{"@id":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/#website"},"primaryImageOfPage":{"@id":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/method\/#primaryimage"},"image":{"@id":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/method\/#primaryimage"},"thumbnailUrl":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/wp-content\/uploads\/sites\/113\/2024\/12\/image.png","datePublished":"2024-05-10T04:55:26+00:00","dateModified":"2024-12-14T04:08:48+00:00","breadcrumb":{"@id":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/method\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/method\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/method\/#primaryimage","url":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/wp-content\/uploads\/sites\/113\/2024\/12\/image.png","contentUrl":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/wp-content\/uploads\/sites\/113\/2024\/12\/image.png","width":713,"height":766},{"@type":"BreadcrumbList","@id":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/method\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/"},{"@type":"ListItem","position":2,"name":"Method"}]},{"@type":"WebSite","@id":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/#website","url":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/","name":"Towards Universal State Estimation and Reconstruction in the Wild","description":"Student:Bhuvan Jhamb, Chenwei Lyu | Advisors: Nikhil Keetha, Yuchen Zhang, Dr. Sebastian Scherer","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/?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\/2024team15\/wp-json\/wp\/v2\/pages\/66","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/wp-json\/wp\/v2\/users\/221"}],"replies":[{"embeddable":true,"href":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/wp-json\/wp\/v2\/comments?post=66"}],"version-history":[{"count":12,"href":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/wp-json\/wp\/v2\/pages\/66\/revisions"}],"predecessor-version":[{"id":305,"href":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/wp-json\/wp\/v2\/pages\/66\/revisions\/305"}],"wp:attachment":[{"href":"https:\/\/mscvprojects.ri.cmu.edu\/2024team15\/wp-json\/wp\/v2\/media?parent=66"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}