detr-resnet-50-finetuned-10k-cppe5

This model is a fine-tuned version of facebook/detr-resnet-50 on the cppe-5 dataset. It achieves the following results on the evaluation set:

• Loss: 0.9865
• Map: 0.3192
• Map 50: 0.6187
• Map 75: 0.3039
• Map small: 0.271
• Map medium: 0.2307
• Map large: 0.4635
• Mar 1: 0.2947
• Mar 10: 0.4873
• Mar 100: 0.5005
• Mar small: 0.3421
• Mar medium: 0.4012
• Mar large: 0.6176
• Map Coverall: 0.5908
• Mar 100 Coverall: 0.6978
• Map Face Shield: 0.3773
• Mar 100 Face Shield: 0.6353
• Map Gloves: 0.2142
• Mar 100 Gloves: 0.3902
• Map Goggles: 0.1483
• Mar 100 Goggles: 0.3906
• Map Mask: 0.2653
• Mar 100 Mask: 0.3880

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 1337
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 100.0
• mixed_precision_training: Native AMP

Framework versions

• Transformers 4.40.0.dev0
• Pytorch 1.13.0+cu117
• Datasets 2.18.0
• Tokenizers 0.19.0