![\"figure](\"https://www.kaggleusercontent.com/kf/218507867/eyJhbGciOiJkaXIiLCJlbmMiOiJBMTI4Q0JDLUhTMjU2In0..X8A_sbwng6grYb4Ff3xORw.jSu0TAkCoVwuhrW_adjEBYdB5fpsA_GiNgcA4Y8tfJ5XiQjsWIPYScsypAHxFequwg10jYFioa0-GXLqr0QCUWtAFdaD3phCl7FGiYPFqkSEvve7FYS_Q3muwGYjCsoICawjqAM7-0O_U_RiVGyOz450e_t7F-OrurcwfGpa5lV-UaEDwlch5kcV1BoTdDdTvx7z6-fljtH8X_MxIAuBP_QnzA_7TQiriy2A7YTzviNuUaDkmvJAxe25ucOjVhtduuobbl4Dugxhr1JuLfJQMqJxnoOJKYwT-KfDQTmz__ZKA-WhEZOXGPHU83yadGOA4bhGei4ga6VhknVUH--ZFHatlTDukyEKBYf4YEJ5skp8woFS1txmyylW0rv5-V51nDS-dQfjrw2_WHi377bea5tkzG7lm6ry8KhjJrBK7gH8--dhM0RPajfHhJAZtb2B0l3TOyzt6XFSsprarGQi8XvJ-UEheXAxZfhmiCNIA9FHHT21RgyYKBwHpQC-rvm6aDF1bob8KrfTnw0y2m77QaHUUCZdo_5pa-a8YDnAb14qIbU2Ux8E00UN_pHylMzlc15JXssiJCjhdlU4_pkkRpaex-_IiYpK3r5XWGLBTXt_E0dAAP_40bv8sLYMyhb0g1qLLjllsB02LnHZMc-aEhhvL_QTknsWXOGgsHd3zo1ZrkzLLZ4246nyVW6HHzoC.w16zKoj8Bo80Ong-gFK15w/__results___files/__results___35_0.png\")
In this competition, the task was to classify images into two categories:
\nTo address this problem, we trained and compared three image classification models from the torchvision.models library:
Vision Transformer (ViT)
\nEfficientNet V2
\nRegNet
\nWe imported the official pretrained weights from torchvision, then replaced the last layer (head or classifier) to output 2 classes (one for the editada class and one for the real class).
Image size:
\nWe standardized all images to 224x224 (or 256x256 with a CenterCrop(224)).
Augmentations:
\nRandomHorizontalFlip, RandomResizedCrop(224), etc. to enrich the dataset and reduce overfitting.[0.485, 0.456, 0.406] / [0.229, 0.224, 0.225]).No complex feature engineering (the features are automatically learned by the CNNs/Transformers).
No manual feature elimination (we used all channels and layers provided by the pretrained weights).
To evaluate our models, we primarily used a train/validation split:
\nIn our final notebooks:
\nAUC on the validation set and the Kaggle public leaderboard.Before making official submissions (to test more strategies, e.g., a “three-way split”):
\nIn the end, the best final validation approach mirroring the final LB was a 95% / 5% split, despite some risk of overfitting.
\nComparing three model families:
\nWe knew that ViT (Vision Transformer), EfficientNet, and RegNet are all competitive for image classification.
Learning rate and Scheduler:
\nlearning rate at 5e-5 and applied a StepLR(gamma=0.7) to reduce the LR every epoch. Data augmentation:
\nChoice of batch size:
\nbatch_size=8 achieved the best results for each model.EfficientNet V2:
\nSmall test-time augmentation (TTA):
\nComparison of a “three-way split” locally:
\nAggressive augmentations (e.g., extreme rotations, strong color jitter):
\nIncreasing input size to 384x384:
\nStacking more final layers or heavier fine-tuning of intermediate layers**:
\nBelow are some sample plots of training and validation curves for each model tested:
\n
![\"figure](\"https://www.kaggleusercontent.com/kf/218507867/eyJhbGciOiJkaXIiLCJlbmMiOiJBMTI4Q0JDLUhTMjU2In0..X8A_sbwng6grYb4Ff3xORw.jSu0TAkCoVwuhrW_adjEBYdB5fpsA_GiNgcA4Y8tfJ5XiQjsWIPYScsypAHxFequwg10jYFioa0-GXLqr0QCUWtAFdaD3phCl7FGiYPFqkSEvve7FYS_Q3muwGYjCsoICawjqAM7-0O_U_RiVGyOz450e_t7F-OrurcwfGpa5lV-UaEDwlch5kcV1BoTdDdTvx7z6-fljtH8X_MxIAuBP_QnzA_7TQiriy2A7YTzviNuUaDkmvJAxe25ucOjVhtduuobbl4Dugxhr1JuLfJQMqJxnoOJKYwT-KfDQTmz__ZKA-WhEZOXGPHU83yadGOA4bhGei4ga6VhknVUH--ZFHatlTDukyEKBYf4YEJ5skp8woFS1txmyylW0rv5-V51nDS-dQfjrw2_WHi377bea5tkzG7lm6ry8KhjJrBK7gH8--dhM0RPajfHhJAZtb2B0l3TOyzt6XFSsprarGQi8XvJ-UEheXAxZfhmiCNIA9FHHT21RgyYKBwHpQC-rvm6aDF1bob8KrfTnw0y2m77QaHUUCZdo_5pa-a8YDnAb14qIbU2Ux8E00UN_pHylMzlc15JXssiJCjhdlU4_pkkRpaex-_IiYpK3r5XWGLBTXt_E0dAAP_40bv8sLYMyhb0g1qLLjllsB02LnHZMc-aEhhvL_QTknsWXOGgsHd3zo1ZrkzLLZ4246nyVW6HHzoC.w16zKoj8Bo80Ong-gFK15w/__results___files/__results___35_1.png\")
![\"figure](\"https://www.kaggleusercontent.com/kf/218364683/eyJhbGciOiJkaXIiLCJlbmMiOiJBMTI4Q0JDLUhTMjU2In0..W4ne8TOi7psjitxhBUJb-w.ztPc5SZr3OT3ZlQ08JLJzzYsEYmQlcvx6c9jwHVSsVS7Z-D1ueftooQ_kqOTlWTrFJE43EHL008Vpnef-XLBrEudqCInmHkY3eTtnTqImmImBloiQyGa8N7MgjT_ufcFwQuWBQN9bdo9_gC6R-Zjp4glHNNbhv_NuHtZXuHEaFqqq_pkk6vaA-xBY8EuBmF_XII_BjelUtD4acwM9fmpbIA3AxWG11qzam5yt22gkDLEB930WG-1Z18qfrJiMhGthfainoNxXfxRVFdvnrf5YNr6xOtjHIdeF_lGK4e8ZTahrL4mRWGdqnpUPvP4I_7XOfytNlpfWL-DB9SzXB822EupaUML_jGh-CkyKnrYYvDbmh57ptQx3dwmFjFKEKuEJbB7ez1Gzh6D_ysYy0LtGwdOdwfpLvW286F7Gt35vL_ATsAOIZ79Qn7cOf0Qbnmo7Q8XP7vvpE7fy7DywVF8LnpkYdqMIuTO9q6pJg1Mi8nZEwkgKkH2EVvIeVHD6p-i-irZmv7Bugcm7xvSSFtEiYG_8HgeBJEusc-vpO4cwxkjYkWG29lCOqjRK285AZ1FhyyXx-LqXqXXlipmZWvWN9BPFQccQh5bfiUNz064JbowLvGGVjGN06cTFvzHKzvT2cSv-idHH5sXRF1kl0wWuw.kCV5oHHuLVRrABAsxe3PFg/__results___files/__results___11_0.png\")
![\"figure](\"https://www.kaggleusercontent.com/kf/218364683/eyJhbGciOiJkaXIiLCJlbmMiOiJBMTI4Q0JDLUhTMjU2In0..W4ne8TOi7psjitxhBUJb-w.ztPc5SZr3OT3ZlQ08JLJzzYsEYmQlcvx6c9jwHVSsVS7Z-D1ueftooQ_kqOTlWTrFJE43EHL008Vpnef-XLBrEudqCInmHkY3eTtnTqImmImBloiQyGa8N7MgjT_ufcFwQuWBQN9bdo9_gC6R-Zjp4glHNNbhv_NuHtZXuHEaFqqq_pkk6vaA-xBY8EuBmF_XII_BjelUtD4acwM9fmpbIA3AxWG11qzam5yt22gkDLEB930WG-1Z18qfrJiMhGthfainoNxXfxRVFdvnrf5YNr6xOtjHIdeF_lGK4e8ZTahrL4mRWGdqnpUPvP4I_7XOfytNlpfWL-DB9SzXB822EupaUML_jGh-CkyKnrYYvDbmh57ptQx3dwmFjFKEKuEJbB7ez1Gzh6D_ysYy0LtGwdOdwfpLvW286F7Gt35vL_ATsAOIZ79Qn7cOf0Qbnmo7Q8XP7vvpE7fy7DywVF8LnpkYdqMIuTO9q6pJg1Mi8nZEwkgKkH2EVvIeVHD6p-i-irZmv7Bugcm7xvSSFtEiYG_8HgeBJEusc-vpO4cwxkjYkWG29lCOqjRK285AZ1FhyyXx-LqXqXXlipmZWvWN9BPFQccQh5bfiUNz064JbowLvGGVjGN06cTFvzHKzvT2cSv-idHH5sXRF1kl0wWuw.kCV5oHHuLVRrABAsxe3PFg/__results___files/__results___11_1.png\")
Finally, I joined the competition only 9 days before its end, so I didn’t really have time to make plenty of submissions. I also chose to respect the Kaggle guideline:
\n\n\n“You cannot sign up to Kaggle from multiple accounts and therefore you cannot enter or submit from multiple accounts.”
\n
…unlike certain others in the Top 10 🥸. Nonetheless, had I been able to iterate a bit more on RegNet (e.g., tuning hyperparameters, adjusting the epoch count), I’m convinced it would have achieved even better performance like +0.95.
\nOne key issue with this competition was the absence of a split between the public and private leaderboard, making brute force submissions a highly effective strategy. By learning from the leaderboard itself, participants could iteratively fine-tune predictions rather than genuinely optimizing the model.
\nThis is apparent in many Top 10 solutions. For instance, bestwater (ranked 1st) openly admits to this approach in their post (link here). Additionally, it's almost certain that solosid69 (top 2, 4, 5, 6, 7) used the same technique—leveraging five different accounts 😶🌫️ to maximize submissions and brute force his way up the leaderboard.
\nIt’s worth noting that bestwater's honesty in admitting this method is commendable, as it underscores the bigger problem: the competition design itself. With no private leaderboard, brute-forcing overshadowed more rigorous model optimization efforts.
\nEven with thorough hyperparameter tuning, the results presented here are likely close to the best achievable without brute-forcing. Nevertheless, I enjoyed this competition a lot; it was a fun and rewarding experience that allowed me to discover new models and gain deeper insights into their performance on real-world data.
\nReference Papers
\nKaggle Notebooks
\nIn this competition, the task was to classify images into two categories:
\nTo address this problem, we trained and compared three image classification models from the torchvision.models library:
Vision Transformer (ViT)
\nEfficientNet V2
\nRegNet
\nWe imported the official pretrained weights from torchvision, then replaced the last layer (head or classifier) to output 2 classes (one for the editada class and one for the real class).
Image size:
\nWe standardized all images to 224x224 (or 256x256 with a CenterCrop(224)).
Augmentations:
\nRandomHorizontalFlip, RandomResizedCrop(224), etc. to enrich the dataset and reduce overfitting.[0.485, 0.456, 0.406] / [0.229, 0.224, 0.225]).No complex feature engineering (the features are automatically learned by the CNNs/Transformers).
No manual feature elimination (we used all channels and layers provided by the pretrained weights).
To evaluate our models, we primarily used a train/validation split:
\nIn our final notebooks:
\nAUC on the validation set and the Kaggle public leaderboard.Before making official submissions (to test more strategies, e.g., a “three-way split”):
\nIn the end, the best final validation approach mirroring the final LB was a 95% / 5% split, despite some risk of overfitting.
\nComparing three model families:
\nWe knew that ViT (Vision Transformer), EfficientNet, and RegNet are all competitive for image classification.
Learning rate and Scheduler:
\nlearning rate at 5e-5 and applied a StepLR(gamma=0.7) to reduce the LR every epoch. Data augmentation:
\nChoice of batch size:
\nbatch_size=8 achieved the best results for each model.EfficientNet V2:
\nSmall test-time augmentation (TTA):
\nComparison of a “three-way split” locally:
\nAggressive augmentations (e.g., extreme rotations, strong color jitter):
\nIncreasing input size to 384x384:
\nStacking more final layers or heavier fine-tuning of intermediate layers**:
\nBelow are some sample plots of training and validation curves for each model tested:
\n
Finally, I joined the competition only 9 days before its end, so I didn’t really have time to make plenty of submissions. I also chose to respect the Kaggle guideline:
\n\n\n“You cannot sign up to Kaggle from multiple accounts and therefore you cannot enter or submit from multiple accounts.”
\n
…unlike certain others in the Top 10 🥸. Nonetheless, had I been able to iterate a bit more on RegNet (e.g., tuning hyperparameters, adjusting the epoch count), I’m convinced it would have achieved even better performance like +0.95.
\nOne key issue with this competition was the absence of a split between the public and private leaderboard, making brute force submissions a highly effective strategy. By learning from the leaderboard itself, participants could iteratively fine-tune predictions rather than genuinely optimizing the model.
\nThis is apparent in many Top 10 solutions. For instance, bestwater (ranked 1st) openly admits to this approach in their post (link here). Additionally, it's almost certain that solosid69 (top 2, 4, 5, 6, 7) used the same technique—leveraging five different accounts 😶🌫️ to maximize submissions and brute force his way up the leaderboard.
\nIt’s worth noting that bestwater's honesty in admitting this method is commendable, as it underscores the bigger problem: the competition design itself. With no private leaderboard, brute-forcing overshadowed more rigorous model optimization efforts.
\nEven with thorough hyperparameter tuning, the results presented here are likely close to the best achievable without brute-forcing. Nevertheless, I enjoyed this competition a lot; it was a fun and rewarding experience that allowed me to discover new models and gain deeper insights into their performance on real-world data.
\nReference Papers
\nKaggle Notebooks
\n