README.md

metadata

library_name: transformers
license: other
base_model: nvidia/segformer-b2-finetuned-cityscapes-1024-1024
tags:
  - vision
  - image-segmentation
  - generated_from_trainer
model-index:
  - name: SegFormer_b2
    results: []

SegFormer_b2

This model is a fine-tuned version of nvidia/segformer-b2-finetuned-cityscapes-1024-1024 on the Cityscapes dataset. It achieves the following results on the evaluation set:

• eval_loss: 0.2516
• eval_mean_iou: 0.3875
• eval_mean_accuracy: 0.5066
• eval_overall_accuracy: 0.9043
• eval_accuracy_unlabeled: nan
• eval_accuracy_ego vehicle: nan
• eval_accuracy_rectification border: nan
• eval_accuracy_out of roi: nan
• eval_accuracy_static: nan
• eval_accuracy_dynamic: nan
• eval_accuracy_ground: nan
• eval_accuracy_road: 0.9832
• eval_accuracy_sidewalk: 0.8421
• eval_accuracy_parking: nan
• eval_accuracy_rail track: nan
• eval_accuracy_building: 0.9158
• eval_accuracy_wall: 0.0
• eval_accuracy_fence: 0.0
• eval_accuracy_guard rail: nan
• eval_accuracy_bridge: nan
• eval_accuracy_tunnel: nan
• eval_accuracy_pole: 0.5362
• eval_accuracy_polegroup: nan
• eval_accuracy_traffic light: 0.5814
• eval_accuracy_traffic sign: 0.7376
• eval_accuracy_vegetation: 0.9188
• eval_accuracy_terrain: 0.6737
• eval_accuracy_sky: 0.9746
• eval_accuracy_person: 0.7788
• eval_accuracy_rider: 0.0
• eval_accuracy_car: 0.9354
• eval_accuracy_truck: 0.0
• eval_accuracy_bus: 0.0
• eval_accuracy_caravan: nan
• eval_accuracy_trailer: nan
• eval_accuracy_train: 0.0
• eval_accuracy_motorcycle: 0.0
• eval_accuracy_bicycle: 0.7472
• eval_accuracy_license plate: nan
• eval_iou_unlabeled: nan
• eval_iou_ego vehicle: nan
• eval_iou_rectification border: nan
• eval_iou_out of roi: nan
• eval_iou_static: 0.0
• eval_iou_dynamic: nan
• eval_iou_ground: nan
• eval_iou_road: 0.9649
• eval_iou_sidewalk: 0.7403
• eval_iou_parking: nan
• eval_iou_rail track: nan
• eval_iou_building: 0.8430
• eval_iou_wall: 0.0
• eval_iou_fence: 0.0
• eval_iou_guard rail: nan
• eval_iou_bridge: nan
• eval_iou_tunnel: nan
• eval_iou_pole: 0.3619
• eval_iou_polegroup: nan
• eval_iou_traffic light: 0.4506
• eval_iou_traffic sign: 0.5317
• eval_iou_vegetation: 0.8647
• eval_iou_terrain: 0.4610
• eval_iou_sky: 0.8806
• eval_iou_person: 0.5967
• eval_iou_rider: 0.0
• eval_iou_car: 0.8756
• eval_iou_truck: 0.0
• eval_iou_bus: 0.0
• eval_iou_caravan: nan
• eval_iou_trailer: nan
• eval_iou_train: 0.0
• eval_iou_motorcycle: 0.0
• eval_iou_bicycle: 0.5665
• eval_iou_license plate: 0.0
• eval_runtime: 185.4692
• eval_samples_per_second: 2.696
• eval_steps_per_second: 0.674
• epoch: 20.4301
• step: 3800

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: 0.0006
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 100
• mixed_precision_training: Native AMP

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0