{"id":13645016,"url":"https://github.com/Rushi314/GPR-Object-Detection","last_synced_at":"2025-04-21T11:32:34.732Z","repository":{"id":134222801,"uuid":"375736810","full_name":"Rushi314/GPR-Object-Detection","owner":"Rushi314","description":"This repository contains code to train object detection models like FRCNN/YOLO for identifying objects in Ground Penetrating Radar scans. It also contains code to generate fake data using Generative Adversarial Networks(GANs). ","archived":false,"fork":false,"pushed_at":"2022-02-01T22:37:15.000Z","size":13244,"stargazers_count":31,"open_issues_count":1,"forks_count":8,"subscribers_count":1,"default_branch":"main","last_synced_at":"2024-11-09T17:42:52.343Z","etag":null,"topics":["frcnn","gans","generative-adversarial-network","ground-penetrating-radar","machine-learning","mmdetection","object-detection","yolo"],"latest_commit_sha":null,"homepage":"","language":"Jupyter Notebook","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/Rushi314.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null}},"created_at":"2021-06-10T15:03:14.000Z","updated_at":"2024-11-02T08:58:18.000Z","dependencies_parsed_at":"2024-01-14T12:28:51.566Z","dependency_job_id":null,"html_url":"https://github.com/Rushi314/GPR-Object-Detection","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Rushi314%2FGPR-Object-Detection","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Rushi314%2FGPR-Object-Detection/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Rushi314%2FGPR-Object-Detection/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Rushi314%2FGPR-Object-Detection/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/Rushi314","download_url":"https://codeload.github.com/Rushi314/GPR-Object-Detection/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":250048017,"owners_count":21366161,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["frcnn","gans","generative-adversarial-network","ground-penetrating-radar","machine-learning","mmdetection","object-detection","yolo"],"created_at":"2024-08-02T01:02:23.711Z","updated_at":"2025-04-21T11:32:34.688Z","avatar_url":"https://github.com/Rushi314.png","language":"Jupyter Notebook","funding_links":[],"categories":["Object Detection Applications"],"sub_categories":[],"readme":"# Detecting Objects in Ground Penetrating Radars Scans\n\nWe used mmdetection library for fine-tuining Faster RCNN, Mask RCNN, and YOLO to \nidentify objects from Ground penetrating radar scans.\n\n### Data: \n- Used publicaly available dataset containing 171 annotated \nGPR scans from https://github.com/irenexychen/gpr-data-classifier.\n- Converted the annotations from .xml format to .coco format using \nhttps://github.com/yukkyo/voc2coco.\n### Creating fake data using GANS: \n- Code: GPR_GAN_071521_v.0.4.ipynb\n#### GANs architecture(created from scratch):\n- Generator: \n  - Input: A tensor of size (100, 1, 1) filled with random numbers from a normal distribution(mean = 0, variance = 1)\n  - Output: A tensor of size (3, 128 , 128) representing Image\n  - NN Architecture: 6 transpose convolutions with batch normalization and ReLU activations. \n- Discriminator:\n  - Input: (3, 128, 128) size tensor either real or fake\n  - Output: Classification of input as either real(1) or fake(0)\n  7 Conv layers with batch normalization and ReLU\n- Loss: Binary Cross Entropy Loss\n\n#### GANs results:\n\n![](GAN_results.png)\n\n### Hyperbola detection:\n- Code: GPR_MMDetection_v0.022.ipynb\n- We finetuned multiple object detection models with various checkpoints and due to the scarsity of GPR scans, \nthe FRCNN model pretrained on MS-COCO dataset with resnet-101\nas backbone gave the best results.\n\n### Results: \n- We finetuned FRCNN with both real images and fake images from GANs\n- Achieved mAP 0.90 in detecting hyperbolas \n\n![](FRCNN_RESULTS.png)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FRushi314%2FGPR-Object-Detection","html_url":"https://awesome.ecosyste.ms/projects/github.com%2FRushi314%2FGPR-Object-Detection","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FRushi314%2FGPR-Object-Detection/lists"}