{
  "id": 362592,
  "title": "58th Place Solution ",
  "url": "/competitions/rsna-2022-cervical-spine-fracture-detection/writeups/arunodhayan-58th-place-solution",
  "author_name": "",
  "post_date": "2022-10-28T00:12:37.369876700Z",
  "votes": 13,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Big thanks to the organizers for the great competition!<br>\nAlso thanks to all competitors for sharing their experiments here</p>\n<h2>Models</h2>\n<p>In this competition, less profound models performed better. I used (EfficientnetB1-B5_ns,EfficientnetV2s,Seresnet101)</p>\n<p>All models trained  on 3 different resolutions - 384<em>384, 515</em>512, 768*768</p>\n<p>5 Fold Cross-validation</p>\n<h2>Augmentations</h2>\n<p>transforms_train = transforms.Compose(<br>\n    [<br>\n        transforms.Resize((IMG_SIZE, IMG_SIZE)),<br>\n        transforms.RandomHorizontalFlip(p=0.3),<br>\n        transforms.RandomVerticalFlip(p=0.3),<br>\n        transforms.RandomResizedCrop(IMG_SIZE),<br>\n        transforms.ToTensor(),<br>\n        transforms.Normalize((0.485, 0.456, 0.406), (0.229, 0.224, 0.225)),<br>\n    ]<br>\n)</p>\n<h2>Best Solution</h2>\n<p>3 Model Ensemble - 0.43 public 0.49 private (Without TTA) - Efficientnet B1-NS Resolution 512 <br>\n9 Model Ensemble - 0.44 public  0.49 private (Without TTA) - Efficientnet B1-NS Resolution 384<br>\n5 Model Ensemble - 0.45 public  0.50 private (Without TTA) - EfficientnetV2s Resolution 384</p>\n<p>3 Model Ensemble - 0.43 public 0.50 private (With TTA) - Efficientnet B1-NS Resolution 512<br>\n9 Model Ensemble - 0.44 public  0.50 private (With TTA) - Efficientnet B1-NS Resolution 384</p>",
  "messages": [
    {
      "id": "2006994",
      "postDate": "10/28/2022 00:12:37",
      "content": "<p>Big thanks to the organizers for the great competition!<br>\nAlso thanks to all competitors for sharing their experiments here</p>\n<h2>Models</h2>\n<p>In this competition, less profound models performed better. I used (EfficientnetB1-B5_ns,EfficientnetV2s,Seresnet101)</p>\n<p>All models trained  on 3 different resolutions - 384<em>384, 515</em>512, 768*768</p>\n<p>5 Fold Cross-validation</p>\n<h2>Augmentations</h2>\n<p>transforms_train = transforms.Compose(<br>\n    [<br>\n        transforms.Resize((IMG_SIZE, IMG_SIZE)),<br>\n        transforms.RandomHorizontalFlip(p=0.3),<br>\n        transforms.RandomVerticalFlip(p=0.3),<br>\n        transforms.RandomResizedCrop(IMG_SIZE),<br>\n        transforms.ToTensor(),<br>\n        transforms.Normalize((0.485, 0.456, 0.406), (0.229, 0.224, 0.225)),<br>\n    ]<br>\n)</p>\n<h2>Best Solution</h2>\n<p>3 Model Ensemble - 0.43 public 0.49 private (Without TTA) - Efficientnet B1-NS Resolution 512 <br>\n9 Model Ensemble - 0.44 public  0.49 private (Without TTA) - Efficientnet B1-NS Resolution 384<br>\n5 Model Ensemble - 0.45 public  0.50 private (Without TTA) - EfficientnetV2s Resolution 384</p>\n<p>3 Model Ensemble - 0.43 public 0.50 private (With TTA) - Efficientnet B1-NS Resolution 512<br>\n9 Model Ensemble - 0.44 public  0.50 private (With TTA) - Efficientnet B1-NS Resolution 384</p>",
      "rawMarkdown": "Big thanks to the organizers for the great competition!\nAlso thanks to all competitors for sharing their experiments here\n##Models\nIn this competition, less profound models performed better. I used (EfficientnetB1-B5_ns,EfficientnetV2s,Seresnet101)\n\nAll models trained  on 3 different resolutions - 384*384, 515*512, 768*768\n\n 5 Fold Cross-validation\n\n##Augmentations\ntransforms_train = transforms.Compose(\n    [\n        transforms.Resize((IMG_SIZE, IMG_SIZE)),\n        transforms.RandomHorizontalFlip(p=0.3),\n        transforms.RandomVerticalFlip(p=0.3),\n        transforms.RandomResizedCrop(IMG_SIZE),\n        transforms.ToTensor(),\n        transforms.Normalize((0.485, 0.456, 0.406), (0.229, 0.224, 0.225)),\n    ]\n)\n\n## Best Solution \n\n3 Model Ensemble - 0.43 public 0.49 private (Without TTA) - Efficientnet B1-NS Resolution 512 \n9 Model Ensemble - 0.44 public  0.49 private (Without TTA) - Efficientnet B1-NS Resolution 384\n5 Model Ensemble - 0.45 public  0.50 private (Without TTA) - EfficientnetV2s Resolution 384\n\n\n3 Model Ensemble - 0.43 public 0.50 private (With TTA) - Efficientnet B1-NS Resolution 512\n9 Model Ensemble - 0.44 public  0.50 private (With TTA) - Efficientnet B1-NS Resolution 384",
      "votes": null
    },
    {
      "id": "2007522",
      "postDate": "10/28/2022 09:36:43",
      "content": "<p>Notebook link <br>\n<a href=\"https://www.kaggle.com/code/arunodhayan/effib1ns/notebook?scriptVersionId=108829887\" target=\"_blank\">https://www.kaggle.com/code/arunodhayan/effib1ns/notebook?scriptVersionId=108829887</a></p>",
      "rawMarkdown": "Notebook link \nhttps://www.kaggle.com/code/arunodhayan/effib1ns/notebook?scriptVersionId=108829887",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2007522,
      "author_name": "arunodhayan",
      "author_url": "",
      "post_date": "10/28/2022 09:36:43",
      "content": "<p>Notebook link <br>\n<a href=\"https://www.kaggle.com/code/arunodhayan/effib1ns/notebook?scriptVersionId=108829887\" target=\"_blank\">https://www.kaggle.com/code/arunodhayan/effib1ns/notebook?scriptVersionId=108829887</a></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2006994": "Big thanks to the organizers for the great competition!\nAlso thanks to all competitors for sharing their experiments here\n##Models\nIn this competition, less profound models performed better. I used (EfficientnetB1-B5_ns,EfficientnetV2s,Seresnet101)\n\nAll models trained  on 3 different resolutions - 384*384, 515*512, 768*768\n\n 5 Fold Cross-validation\n\n##Augmentations\ntransforms_train = transforms.Compose(\n    [\n        transforms.Resize((IMG_SIZE, IMG_SIZE)),\n        transforms.RandomHorizontalFlip(p=0.3),\n        transforms.RandomVerticalFlip(p=0.3),\n        transforms.RandomResizedCrop(IMG_SIZE),\n        transforms.ToTensor(),\n        transforms.Normalize((0.485, 0.456, 0.406), (0.229, 0.224, 0.225)),\n    ]\n)\n\n## Best Solution \n\n3 Model Ensemble - 0.43 public 0.49 private (Without TTA) - Efficientnet B1-NS Resolution 512 \n9 Model Ensemble - 0.44 public  0.49 private (Without TTA) - Efficientnet B1-NS Resolution 384\n5 Model Ensemble - 0.45 public  0.50 private (Without TTA) - EfficientnetV2s Resolution 384\n\n\n3 Model Ensemble - 0.43 public 0.50 private (With TTA) - Efficientnet B1-NS Resolution 512\n9 Model Ensemble - 0.44 public  0.50 private (With TTA) - Efficientnet B1-NS Resolution 384",
    "2007522": "Notebook link \nhttps://www.kaggle.com/code/arunodhayan/effib1ns/notebook?scriptVersionId=108829887"
  },
  "source": "meta"
}