segformer-b0-finetuned-segments-sidewalk-oct-22

This model is a fine-tuned version of nvidia/mit-b0 on the segments/sidewalk-semantic dataset. It achieves the following results on the evaluation set:

• eval_loss: 1.2015
• eval_mean_iou: 0.1448
• eval_mean_accuracy: 0.1852
• eval_overall_accuracy: 0.7367
• eval_accuracy_unlabeled: nan
• eval_accuracy_flat-road: 0.7598
• eval_accuracy_flat-sidewalk: 0.9596
• eval_accuracy_flat-crosswalk: 0.0001
• eval_accuracy_flat-cyclinglane: 0.3703
• eval_accuracy_flat-parkingdriveway: 0.0018
• eval_accuracy_flat-railtrack: 0.0
• eval_accuracy_flat-curb: 0.0
• eval_accuracy_human-person: 0.0
• eval_accuracy_human-rider: 0.0
• eval_accuracy_vehicle-car: 0.8814
• eval_accuracy_vehicle-truck: 0.0
• eval_accuracy_vehicle-bus: 0.0
• eval_accuracy_vehicle-tramtrain: 0.0
• eval_accuracy_vehicle-motorcycle: 0.0
• eval_accuracy_vehicle-bicycle: 0.0
• eval_accuracy_vehicle-caravan: 0.0
• eval_accuracy_vehicle-cartrailer: 0.0
• eval_accuracy_construction-building: 0.8832
• eval_accuracy_construction-door: 0.0
• eval_accuracy_construction-wall: 0.0
• eval_accuracy_construction-fenceguardrail: 0.0
• eval_accuracy_construction-bridge: 0.0
• eval_accuracy_construction-tunnel: 0.0
• eval_accuracy_construction-stairs: 0.0
• eval_accuracy_object-pole: 0.0
• eval_accuracy_object-trafficsign: 0.0
• eval_accuracy_object-trafficlight: 0.0
• eval_accuracy_nature-vegetation: 0.9387
• eval_accuracy_nature-terrain: 0.6141
• eval_accuracy_sky: 0.8885
• eval_accuracy_void-ground: 0.0
• eval_accuracy_void-dynamic: 0.0
• eval_accuracy_void-static: 0.0
• eval_accuracy_void-unclear: 0.0
• eval_iou_unlabeled: nan
• eval_iou_flat-road: 0.5640
• eval_iou_flat-sidewalk: 0.7376
• eval_iou_flat-crosswalk: 0.0001
• eval_iou_flat-cyclinglane: 0.3544
• eval_iou_flat-parkingdriveway: 0.0018
• eval_iou_flat-railtrack: 0.0
• eval_iou_flat-curb: 0.0
• eval_iou_human-person: 0.0
• eval_iou_human-rider: 0.0
• eval_iou_vehicle-car: 0.5854
• eval_iou_vehicle-truck: 0.0
• eval_iou_vehicle-bus: 0.0
• eval_iou_vehicle-tramtrain: 0.0
• eval_iou_vehicle-motorcycle: 0.0
• eval_iou_vehicle-bicycle: 0.0
• eval_iou_vehicle-caravan: 0.0
• eval_iou_vehicle-cartrailer: 0.0
• eval_iou_construction-building: 0.5548
• eval_iou_construction-door: 0.0
• eval_iou_construction-wall: 0.0
• eval_iou_construction-fenceguardrail: 0.0
• eval_iou_construction-bridge: 0.0
• eval_iou_construction-tunnel: 0.0
• eval_iou_construction-stairs: 0.0
• eval_iou_object-pole: 0.0
• eval_iou_object-trafficsign: 0.0
• eval_iou_object-trafficlight: 0.0
• eval_iou_nature-vegetation: 0.7404
• eval_iou_nature-terrain: 0.5555
• eval_iou_sky: 0.8277
• eval_iou_void-ground: 0.0
• eval_iou_void-dynamic: 0.0
• eval_iou_void-static: 0.0
• eval_iou_void-unclear: 0.0
• eval_runtime: 35.456
• eval_samples_per_second: 5.641
• eval_steps_per_second: 0.282
• epoch: 0.5
• step: 20

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: 20
• eval_batch_size: 20
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Framework versions

• Transformers 4.38.2
• Pytorch 2.1.0+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2