说明文档

图片分级（safe、r15、r18）预测模型。

请注意，由于 safe、r15 和 r18 等级之间的界限通常不明确，此任务不存在客观的基准真值，且数据未经人工筛选直接爬取获得。因此，基于此数据集训练的模型只能提供粗略的判断。如果您需要对 R18 图片进行精确分类，建议考虑基于关键点目标检测的解决方案。

模型	FLOPs	准确率	混淆矩阵	描述
caformer_s36_plus	22.10G	74.26%	混淆矩阵	模型：caformer_s36.sail_in22k_ft_in1k_384，timm 预训练
mobilenetv3	0.63G	64.77%	混淆矩阵	模型：mobilenetv3_large_100，来自 timm
mobilenetv3_sce	0.63G	66.27%	混淆矩阵	模型：mobilenetv3_large_100，来自 timm，使用 SCELoss 作为损失函数
mobilenetv3_sce_dist	0.63G	69.49%	混淆矩阵	从 caformer_s36_plus 蒸馏而来，使用 mobilenetv3_large_100

deepghs/anime_rating

作者 deepghs

image-classification

↓ 0 ♥ 3

创建时间: 2023-06-03 06:24:55+00:00

更新时间: 2024-01-19 15:45:52+00:00

在 Hugging Face 上查看

文件 (80)

.gitattributes

README.md

caformer_b36_v1_pruned_ls0.1/meta.json

caformer_b36_v1_pruned_ls0.1/metrics.json

caformer_b36_v1_pruned_ls0.1/model.ckpt

caformer_b36_v1_pruned_ls0.1/model.onnx ONNX

caformer_b36_v1_pruned_ls0.1/plot_confusion.png

caformer_b36_v1_pruned_ls0.1/plot_f1_curve.png

caformer_b36_v1_pruned_ls0.1/plot_p_curve.png

caformer_b36_v1_pruned_ls0.1/plot_pr_curve.png

caformer_b36_v1_pruned_ls0.1/plot_r_curve.png

caformer_b36_v1_pruned_ls0.1/plot_roc_curve.png

caformer_b36_v1_pruned_ls0.1/plot_sample_r15.png

caformer_b36_v1_pruned_ls0.1/plot_sample_r18.png

caformer_b36_v1_pruned_ls0.1/plot_sample_safe.png

caformer_s36_plus/meta.json

caformer_s36_plus/metrics.json

caformer_s36_plus/model.ckpt

caformer_s36_plus/model.onnx ONNX

caformer_s36_plus/plot_confusion.png

caformer_s36_plus/plot_f1_curve.png

caformer_s36_plus/plot_p_curve.png

caformer_s36_plus/plot_pr_curve.png

caformer_s36_plus/plot_r_curve.png

caformer_s36_plus/plot_roc_curve.png

caformer_s36_v1_pruned_ls0.1/meta.json

caformer_s36_v1_pruned_ls0.1/metrics.json

caformer_s36_v1_pruned_ls0.1/model.ckpt

caformer_s36_v1_pruned_ls0.1/model.onnx ONNX

caformer_s36_v1_pruned_ls0.1/plot_confusion.png

caformer_s36_v1_pruned_ls0.1/plot_f1_curve.png

caformer_s36_v1_pruned_ls0.1/plot_p_curve.png

caformer_s36_v1_pruned_ls0.1/plot_pr_curve.png

caformer_s36_v1_pruned_ls0.1/plot_r_curve.png

caformer_s36_v1_pruned_ls0.1/plot_roc_curve.png

caformer_s36_v1_pruned_ls0.1/plot_sample_r15.png

caformer_s36_v1_pruned_ls0.1/plot_sample_r18.png

caformer_s36_v1_pruned_ls0.1/plot_sample_safe.png

mobilenetv3/meta.json

mobilenetv3/metrics.json

mobilenetv3/model.ckpt

mobilenetv3/model.onnx ONNX

mobilenetv3/plot_confusion.png

mobilenetv3/plot_f1_curve.png

mobilenetv3/plot_p_curve.png

mobilenetv3/plot_pr_curve.png

mobilenetv3/plot_r_curve.png

mobilenetv3_sce/meta.json

mobilenetv3_sce/metrics.json

mobilenetv3_sce/model.ckpt

mobilenetv3_sce/model.onnx ONNX

mobilenetv3_sce/plot_confusion.png

mobilenetv3_sce/plot_f1_curve.png

mobilenetv3_sce/plot_p_curve.png

mobilenetv3_sce/plot_pr_curve.png

mobilenetv3_sce/plot_r_curve.png

mobilenetv3_sce/plot_roc_curve.png

mobilenetv3_sce_dist/meta.json

mobilenetv3_sce_dist/metrics.json

mobilenetv3_sce_dist/model.ckpt

mobilenetv3_sce_dist/model.onnx ONNX

mobilenetv3_sce_dist/plot_confusion.png

mobilenetv3_sce_dist/plot_f1_curve.png

mobilenetv3_sce_dist/plot_p_curve.png

mobilenetv3_sce_dist/plot_pr_curve.png

mobilenetv3_sce_dist/plot_r_curve.png

mobilenetv3_sce_dist/plot_roc_curve.png

mobilenetv3_v1_pruned_ls0.1/meta.json

mobilenetv3_v1_pruned_ls0.1/metrics.json

mobilenetv3_v1_pruned_ls0.1/model.ckpt

mobilenetv3_v1_pruned_ls0.1/model.onnx ONNX

mobilenetv3_v1_pruned_ls0.1/plot_confusion.png

mobilenetv3_v1_pruned_ls0.1/plot_f1_curve.png

mobilenetv3_v1_pruned_ls0.1/plot_p_curve.png

mobilenetv3_v1_pruned_ls0.1/plot_pr_curve.png

mobilenetv3_v1_pruned_ls0.1/plot_r_curve.png

mobilenetv3_v1_pruned_ls0.1/plot_roc_curve.png

mobilenetv3_v1_pruned_ls0.1/plot_sample_r15.png

mobilenetv3_v1_pruned_ls0.1/plot_sample_r18.png

mobilenetv3_v1_pruned_ls0.1/plot_sample_safe.png