{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/efficient-yet-deep-convolutional-neural","title":"Efficient Yet Deep Convolutional Neural Networks for Semantic Segmentation","arxiv_id":"1707.08254","date":"2017-07-26","proceeding":null,"authors":["Sharif Amit Kamran","Ali Shihab Sabbir"],"abstract":"Semantic Segmentation using deep convolutional neural network pose more\ncomplex challenge for any GPU intensive task. As it has to compute million of\nparameters, it results to huge memory consumption. Moreover, extracting finer\nfeatures and conducting supervised training tends to increase the complexity.\nWith the introduction of Fully Convolutional Neural Network, which uses finer\nstrides and utilizes deconvolutional layers for upsampling, it has been a go to\nfor any image segmentation task. In this paper, we propose two segmentation\narchitecture which not only needs one-third the parameters to compute but also\ngives better accuracy than the similar architectures. The model weights were\ntransferred from the popular neural net like VGG19 and VGG16 which were trained\non Imagenet classification data-set. Then we transform all the fully connected\nlayers to convolutional layers and use dilated convolution for decreasing the\nparameters. Lastly, we add finer strides and attach four skip architectures\nwhich are element-wise summed with the deconvolutional layers in steps. We\ntrain and test on different sparse and fine data-sets like Pascal VOC2012,\nPascal-Context and NYUDv2 and show how better our model performs in this tasks.\nOn the other hand our model has a faster inference time and consumes less\nmemory for training and testing on NVIDIA Pascal GPUs, making it more efficient\nand less memory consuming architecture for pixel-wise segmentation.","url_abs":"http://arxiv.org/abs/1707.08254v3","url_pdf":"http://arxiv.org/pdf/1707.08254v3.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"efficient-yet-deep-convolutional-neural","repo_url":"https://github.com/SharifAmit/DilatedFCNSegmentation","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"caffe2","reach":null}],"tasks":[{"task_slug":null,"task_name":"GPU"},{"task_slug":"image-segmentation","task_name":"Image Segmentation"},{"task_slug":"scene-segmentation","task_name":"Scene Segmentation"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"dilated-convolution","method_name":"Dilated Convolution"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/scene-segmentation-on-nyu-depth-v2","task":"Scene Segmentation","dataset":"NYU Depth v2","model":"Dilated FCN-2s RGB","rank_in_archive_order":1,"of":1,"metrics":{"Mean IoU":"32.3%"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-nyu-depth-v2","task":"Semantic Segmentation","dataset":"NYU Depth v2","model":"Dilated FCN-2s RGB","rank_in_archive_order":117,"of":121,"metrics":{"Mean IoU":"32.3%"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-pascal-context","task":"Semantic Segmentation","dataset":"PASCAL Context","model":"Dilated-FCN2s","rank_in_archive_order":58,"of":66,"metrics":{"mIoU":"42.6"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-pascal-voc-2012","task":"Semantic Segmentation","dataset":"PASCAL VOC 2012 test","model":"Dilated FCN-2s VGG19","rank_in_archive_order":40,"of":51,"metrics":{"Mean IoU":"69%"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}