Instructions to use james-emi/darpa_ir-f8v2-person-mask2former-swinb-v4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use james-emi/darpa_ir-f8v2-person-mask2former-swinb-v4 with Transformers:
# Load model directly from transformers import AutoImageProcessor, Mask2FormerForUniversalSegmentation processor = AutoImageProcessor.from_pretrained("james-emi/darpa_ir-f8v2-person-mask2former-swinb-v4") model = Mask2FormerForUniversalSegmentation.from_pretrained("james-emi/darpa_ir-f8v2-person-mask2former-swinb-v4") - Notebooks
- Google Colab
- Kaggle
Training in progress, epoch 1
Browse files- config.json +105 -0
- model.safetensors +3 -0
- preprocessor_config.json +27 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,105 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation_function": "relu",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"Mask2FormerForUniversalSegmentation"
|
| 5 |
+
],
|
| 6 |
+
"backbone": null,
|
| 7 |
+
"backbone_config": {
|
| 8 |
+
"attention_probs_dropout_prob": 0.0,
|
| 9 |
+
"depths": [
|
| 10 |
+
2,
|
| 11 |
+
2,
|
| 12 |
+
18,
|
| 13 |
+
2
|
| 14 |
+
],
|
| 15 |
+
"drop_path_rate": 0.3,
|
| 16 |
+
"embed_dim": 128,
|
| 17 |
+
"encoder_stride": 32,
|
| 18 |
+
"hidden_act": "gelu",
|
| 19 |
+
"hidden_dropout_prob": 0.0,
|
| 20 |
+
"hidden_size": 1024,
|
| 21 |
+
"image_size": 224,
|
| 22 |
+
"initializer_range": 0.02,
|
| 23 |
+
"layer_norm_eps": 1e-05,
|
| 24 |
+
"mlp_ratio": 4.0,
|
| 25 |
+
"model_type": "swin",
|
| 26 |
+
"num_channels": 3,
|
| 27 |
+
"num_heads": [
|
| 28 |
+
4,
|
| 29 |
+
8,
|
| 30 |
+
16,
|
| 31 |
+
32
|
| 32 |
+
],
|
| 33 |
+
"num_layers": 4,
|
| 34 |
+
"out_features": [
|
| 35 |
+
"stage1",
|
| 36 |
+
"stage2",
|
| 37 |
+
"stage3",
|
| 38 |
+
"stage4"
|
| 39 |
+
],
|
| 40 |
+
"out_indices": [
|
| 41 |
+
1,
|
| 42 |
+
2,
|
| 43 |
+
3,
|
| 44 |
+
4
|
| 45 |
+
],
|
| 46 |
+
"patch_size": 4,
|
| 47 |
+
"path_norm": true,
|
| 48 |
+
"qkv_bias": true,
|
| 49 |
+
"stage_names": [
|
| 50 |
+
"stem",
|
| 51 |
+
"stage1",
|
| 52 |
+
"stage2",
|
| 53 |
+
"stage3",
|
| 54 |
+
"stage4"
|
| 55 |
+
],
|
| 56 |
+
"torch_dtype": "float32",
|
| 57 |
+
"use_absolute_embeddings": false,
|
| 58 |
+
"window_size": 12
|
| 59 |
+
},
|
| 60 |
+
"backbone_kwargs": null,
|
| 61 |
+
"class_weight": 2.0,
|
| 62 |
+
"common_stride": 4,
|
| 63 |
+
"decoder_layers": 10,
|
| 64 |
+
"dice_weight": 5.0,
|
| 65 |
+
"dim_feedforward": 2048,
|
| 66 |
+
"dropout": 0.0,
|
| 67 |
+
"encoder_feedforward_dim": 1024,
|
| 68 |
+
"encoder_layers": 6,
|
| 69 |
+
"enforce_input_proj": false,
|
| 70 |
+
"enforce_input_projection": false,
|
| 71 |
+
"feature_size": 256,
|
| 72 |
+
"feature_strides": [
|
| 73 |
+
4,
|
| 74 |
+
8,
|
| 75 |
+
16,
|
| 76 |
+
32
|
| 77 |
+
],
|
| 78 |
+
"hidden_dim": 256,
|
| 79 |
+
"id2label": {
|
| 80 |
+
"0": "person"
|
| 81 |
+
},
|
| 82 |
+
"ignore_value": 255,
|
| 83 |
+
"importance_sample_ratio": 0.75,
|
| 84 |
+
"init_std": 0.02,
|
| 85 |
+
"init_xavier_std": 1.0,
|
| 86 |
+
"label2id": {
|
| 87 |
+
"person": 0
|
| 88 |
+
},
|
| 89 |
+
"mask_feature_size": 256,
|
| 90 |
+
"mask_weight": 5.0,
|
| 91 |
+
"model_type": "mask2former",
|
| 92 |
+
"no_object_weight": 0.1,
|
| 93 |
+
"num_attention_heads": 8,
|
| 94 |
+
"num_hidden_layers": 10,
|
| 95 |
+
"num_queries": 100,
|
| 96 |
+
"output_auxiliary_logits": null,
|
| 97 |
+
"oversample_ratio": 3.0,
|
| 98 |
+
"pre_norm": false,
|
| 99 |
+
"torch_dtype": "float32",
|
| 100 |
+
"train_num_points": 12544,
|
| 101 |
+
"transformers_version": "4.55.1",
|
| 102 |
+
"use_auxiliary_loss": true,
|
| 103 |
+
"use_pretrained_backbone": false,
|
| 104 |
+
"use_timm_backbone": false
|
| 105 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9ea2029c99889c4d426bd5b3f4c75f63ce7e9e8514fa3d2a00390420429a1d34
|
| 3 |
+
size 431631824
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_normalize": true,
|
| 3 |
+
"do_reduce_labels": true,
|
| 4 |
+
"do_rescale": true,
|
| 5 |
+
"do_resize": true,
|
| 6 |
+
"ignore_index": 255,
|
| 7 |
+
"image_mean": [
|
| 8 |
+
0.48500001430511475,
|
| 9 |
+
0.4560000002384186,
|
| 10 |
+
0.4059999883174896
|
| 11 |
+
],
|
| 12 |
+
"image_processor_type": "Mask2FormerImageProcessor",
|
| 13 |
+
"image_std": [
|
| 14 |
+
0.2290000021457672,
|
| 15 |
+
0.2239999920129776,
|
| 16 |
+
0.22499999403953552
|
| 17 |
+
],
|
| 18 |
+
"num_labels": 80,
|
| 19 |
+
"pad_size": null,
|
| 20 |
+
"resample": 2,
|
| 21 |
+
"rescale_factor": 0.00392156862745098,
|
| 22 |
+
"size": {
|
| 23 |
+
"height": 481,
|
| 24 |
+
"width": 641
|
| 25 |
+
},
|
| 26 |
+
"size_divisor": 32
|
| 27 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fa9b26a39e255099c063852f544a250019fd35fa2085d15df7d4faeaada94578
|
| 3 |
+
size 5777
|