--- license: other tags: - generated_from_trainer model-index: - name: segformer-b0-finetuned-segments-rowbody-4cats results: [] --- # segformer-b0-finetuned-segments-rowbody-4cats This model is a fine-tuned version of [nvidia/mit-b0](https://huggingface.co/nvidia/mit-b0) on the None dataset. It achieves the following results on the evaluation set: - Loss: 0.1486 - Mean Iou: 0.6527 - Mean Accuracy: 0.9381 - Overall Accuracy: 0.9558 - Accuracy Sleeve-right: nan - Accuracy Sleeve-left: 0.9259 - Accuracy Neck: 0.9212 - Accuracy Body: 0.9670 - Iou Sleeve-right: 0.0 - Iou Sleeve-left: 0.9012 - Iou Neck: 0.7545 - Iou Body: 0.9551 ## 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: 6e-05 - train_batch_size: 1 - eval_batch_size: 1 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 50 ### Training results | Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Sleeve-right | Accuracy Sleeve-left | Accuracy Neck | Accuracy Body | Iou Sleeve-right | Iou Sleeve-left | Iou Neck | Iou Body | |:-------------:|:-----:|:----:|:---------------:|:--------:|:-------------:|:----------------:|:---------------------:|:--------------------:|:-------------:|:-------------:|:----------------:|:---------------:|:--------:|:--------:| | 0.9629 | 2.5 | 20 | 1.1228 | 0.2325 | 0.3900 | 0.7680 | nan | 0.1708 | 0.0 | 0.9991 | 0.0 | 0.1703 | 0.0 | 0.7598 | | 0.6667 | 5.0 | 40 | 0.6465 | 0.4513 | 0.6646 | 0.8681 | nan | 0.5731 | 0.4337 | 0.9870 | 0.0 | 0.5659 | 0.3911 | 0.8483 | | 0.414 | 7.5 | 60 | 0.4340 | 0.5935 | 0.8506 | 0.9377 | nan | 0.8496 | 0.7215 | 0.9806 | 0.0 | 0.8374 | 0.6136 | 0.9228 | | 0.3351 | 10.0 | 80 | 0.3371 | 0.6263 | 0.9237 | 0.9411 | nan | 0.9325 | 0.8909 | 0.9478 | 0.0 | 0.8568 | 0.7171 | 0.9314 | | 0.4849 | 12.5 | 100 | 0.3146 | 0.6428 | 0.9438 | 0.9526 | nan | 0.9708 | 0.9093 | 0.9513 | 0.0 | 0.8922 | 0.7375 | 0.9415 | | 0.2461 | 15.0 | 120 | 0.2716 | 0.6431 | 0.9554 | 0.9496 | nan | 0.9717 | 0.9511 | 0.9434 | 0.0 | 0.8938 | 0.7410 | 0.9373 | | 0.2123 | 17.5 | 140 | 0.2477 | 0.6715 | 0.9481 | 0.9679 | nan | 0.9523 | 0.9151 | 0.9768 | 0.0 | 0.9229 | 0.7999 | 0.9632 | | 0.1827 | 20.0 | 160 | 0.2413 | 0.6620 | 0.9048 | 0.9649 | nan | 0.9200 | 0.8032 | 0.9912 | 0.0 | 0.9147 | 0.7717 | 0.9617 | | 0.2828 | 22.5 | 180 | 0.2286 | 0.6484 | 0.9472 | 0.9532 | nan | 0.9622 | 0.9265 | 0.9530 | 0.0 | 0.8996 | 0.7495 | 0.9443 | | 0.4631 | 25.0 | 200 | 0.2137 | 0.6459 | 0.9452 | 0.9485 | nan | 0.9523 | 0.9345 | 0.9486 | 0.0 | 0.8886 | 0.7543 | 0.9408 | | 0.159 | 27.5 | 220 | 0.1854 | 0.6336 | 0.9374 | 0.9415 | nan | 0.9355 | 0.9328 | 0.9440 | 0.0 | 0.8832 | 0.7145 | 0.9368 | | 0.1361 | 30.0 | 240 | 0.1760 | 0.6563 | 0.9429 | 0.9576 | nan | 0.9492 | 0.9161 | 0.9635 | 0.0 | 0.9027 | 0.7687 | 0.9538 | | 0.1369 | 32.5 | 260 | 0.1634 | 0.6516 | 0.9451 | 0.9562 | nan | 0.9497 | 0.9249 | 0.9607 | 0.0 | 0.9130 | 0.7426 | 0.9507 | | 0.1356 | 35.0 | 280 | 0.1580 | 0.6488 | 0.9248 | 0.9527 | nan | 0.9154 | 0.8905 | 0.9684 | 0.0 | 0.8966 | 0.7457 | 0.9528 | | 0.2415 | 37.5 | 300 | 0.1635 | 0.6457 | 0.9439 | 0.9523 | nan | 0.9406 | 0.9340 | 0.9571 | 0.0 | 0.9003 | 0.7344 | 0.9482 | | 0.1183 | 40.0 | 320 | 0.1639 | 0.6463 | 0.9386 | 0.9524 | nan | 0.9286 | 0.9262 | 0.9612 | 0.0 | 0.8998 | 0.7351 | 0.9500 | | 0.1264 | 42.5 | 340 | 0.1564 | 0.6512 | 0.9439 | 0.9552 | nan | 0.9385 | 0.9312 | 0.9618 | 0.0 | 0.9066 | 0.7463 | 0.9521 | | 0.2418 | 45.0 | 360 | 0.1553 | 0.6569 | 0.9366 | 0.9589 | nan | 0.9198 | 0.9166 | 0.9734 | 0.0 | 0.9033 | 0.7656 | 0.9587 | | 0.189 | 47.5 | 380 | 0.1611 | 0.6520 | 0.9386 | 0.9544 | nan | 0.9284 | 0.9231 | 0.9643 | 0.0 | 0.8990 | 0.7554 | 0.9536 | | 0.1132 | 50.0 | 400 | 0.1486 | 0.6527 | 0.9381 | 0.9558 | nan | 0.9259 | 0.9212 | 0.9670 | 0.0 | 0.9012 | 0.7545 | 0.9551 | ### Framework versions - Transformers 4.25.1 - Pytorch 1.13.0+cu116 - Datasets 2.8.0 - Tokenizers 0.13.2