upload fine-tuned RT-DETRv2 trashify object detection model
Browse files- README.md +95 -0
- config.json +136 -0
- model.safetensors +3 -0
- training_args.bin +3 -0
README.md
ADDED
@@ -0,0 +1,95 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
library_name: transformers
|
3 |
+
license: apache-2.0
|
4 |
+
base_model: PekingU/rtdetr_v2_r50vd
|
5 |
+
tags:
|
6 |
+
- generated_from_trainer
|
7 |
+
model-index:
|
8 |
+
- name: rt_detrv2_finetuned_trashify_box_detector_v1
|
9 |
+
results: []
|
10 |
+
---
|
11 |
+
|
12 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
13 |
+
should probably proofread and complete it, then remove this comment. -->
|
14 |
+
|
15 |
+
# rt_detrv2_finetuned_trashify_box_detector_v1
|
16 |
+
|
17 |
+
This model is a fine-tuned version of [PekingU/rtdetr_v2_r50vd](https://huggingface.co/PekingU/rtdetr_v2_r50vd) on an unknown dataset.
|
18 |
+
It achieves the following results on the evaluation set:
|
19 |
+
- Loss: 9.6943
|
20 |
+
- Map: 0.5019
|
21 |
+
- Map 50: 0.6587
|
22 |
+
- Map 75: 0.5594
|
23 |
+
- Map Small: 0.2
|
24 |
+
- Map Medium: 0.1466
|
25 |
+
- Map Large: 0.5289
|
26 |
+
- Mar 1: 0.5637
|
27 |
+
- Mar 10: 0.7214
|
28 |
+
- Mar 100: 0.7844
|
29 |
+
- Mar Small: 0.4
|
30 |
+
- Mar Medium: 0.5023
|
31 |
+
- Mar Large: 0.826
|
32 |
+
- Map Bin: 0.7483
|
33 |
+
- Mar 100 Bin: 0.9
|
34 |
+
- Map Hand: 0.5688
|
35 |
+
- Mar 100 Hand: 0.8225
|
36 |
+
- Map Not Bin: 0.1236
|
37 |
+
- Mar 100 Not Bin: 0.6357
|
38 |
+
- Map Not Hand: -1.0
|
39 |
+
- Mar 100 Not Hand: -1.0
|
40 |
+
- Map Not Trash: 0.2696
|
41 |
+
- Mar 100 Not Trash: 0.6236
|
42 |
+
- Map Trash: 0.668
|
43 |
+
- Mar 100 Trash: 0.7912
|
44 |
+
- Map Trash Arm: 0.6334
|
45 |
+
- Mar 100 Trash Arm: 0.9333
|
46 |
+
|
47 |
+
## Model description
|
48 |
+
|
49 |
+
More information needed
|
50 |
+
|
51 |
+
## Intended uses & limitations
|
52 |
+
|
53 |
+
More information needed
|
54 |
+
|
55 |
+
## Training and evaluation data
|
56 |
+
|
57 |
+
More information needed
|
58 |
+
|
59 |
+
## Training procedure
|
60 |
+
|
61 |
+
### Training hyperparameters
|
62 |
+
|
63 |
+
The following hyperparameters were used during training:
|
64 |
+
- learning_rate: 0.0001
|
65 |
+
- train_batch_size: 16
|
66 |
+
- eval_batch_size: 16
|
67 |
+
- seed: 42
|
68 |
+
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
|
69 |
+
- lr_scheduler_type: linear
|
70 |
+
- lr_scheduler_warmup_ratio: 0.05
|
71 |
+
- num_epochs: 10
|
72 |
+
- mixed_precision_training: Native AMP
|
73 |
+
|
74 |
+
### Training results
|
75 |
+
|
76 |
+
| Training Loss | Epoch | Step | Validation Loss | Map | Map 50 | Map 75 | Map Small | Map Medium | Map Large | Mar 1 | Mar 10 | Mar 100 | Mar Small | Mar Medium | Mar Large | Map Bin | Mar 100 Bin | Map Hand | Mar 100 Hand | Map Not Bin | Mar 100 Not Bin | Map Not Hand | Mar 100 Not Hand | Map Not Trash | Mar 100 Not Trash | Map Trash | Mar 100 Trash | Map Trash Arm | Mar 100 Trash Arm |
|
77 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|:------:|:---------:|:----------:|:---------:|:------:|:------:|:-------:|:---------:|:----------:|:---------:|:-------:|:-----------:|:--------:|:------------:|:-----------:|:---------------:|:------------:|:----------------:|:-------------:|:-----------------:|:---------:|:-------------:|:-------------:|:-----------------:|
|
78 |
+
| 95.4203 | 1.0 | 50 | 36.7632 | 0.0969 | 0.1711 | 0.0953 | 0.0 | 0.0279 | 0.099 | 0.154 | 0.2806 | 0.2876 | 0.0 | 0.1466 | 0.3008 | 0.3025 | 0.5014 | 0.1779 | 0.4696 | 0.0005 | 0.0714 | -1.0 | -1.0 | 0.0045 | 0.1847 | 0.0945 | 0.3319 | 0.0012 | 0.1667 |
|
79 |
+
| 34.5234 | 2.0 | 100 | 17.3569 | 0.2897 | 0.4017 | 0.3215 | 0.0 | 0.0752 | 0.299 | 0.3142 | 0.5754 | 0.6584 | 0.0 | 0.2153 | 0.7096 | 0.6262 | 0.8298 | 0.5167 | 0.7873 | 0.0294 | 0.4286 | -1.0 | -1.0 | 0.0495 | 0.4819 | 0.5141 | 0.6894 | 0.0021 | 0.7333 |
|
80 |
+
| 21.3497 | 3.0 | 150 | 12.3189 | 0.343 | 0.4639 | 0.3804 | 0.0 | 0.1743 | 0.3651 | 0.4478 | 0.6271 | 0.7397 | 0.0 | 0.4795 | 0.7863 | 0.6933 | 0.8738 | 0.5149 | 0.8304 | 0.0087 | 0.5786 | -1.0 | -1.0 | 0.1921 | 0.5917 | 0.6081 | 0.7637 | 0.0409 | 0.8 |
|
81 |
+
| 17.6468 | 4.0 | 200 | 10.5626 | 0.439 | 0.5844 | 0.5068 | 0.0875 | 0.1241 | 0.4618 | 0.5399 | 0.6942 | 0.7454 | 0.35 | 0.5125 | 0.7766 | 0.7371 | 0.8936 | 0.5741 | 0.8284 | 0.0357 | 0.6 | -1.0 | -1.0 | 0.2263 | 0.6167 | 0.6307 | 0.7673 | 0.43 | 0.7667 |
|
82 |
+
| 15.7918 | 5.0 | 250 | 10.4776 | 0.4151 | 0.5608 | 0.4545 | 0.175 | 0.1627 | 0.4374 | 0.5173 | 0.6789 | 0.7629 | 0.35 | 0.4608 | 0.8144 | 0.7314 | 0.8986 | 0.4641 | 0.8196 | 0.019 | 0.5643 | -1.0 | -1.0 | 0.2707 | 0.6625 | 0.6599 | 0.7991 | 0.3455 | 0.8333 |
|
83 |
+
| 14.5311 | 6.0 | 300 | 10.0502 | 0.4165 | 0.5597 | 0.4713 | 0.12 | 0.1546 | 0.445 | 0.5331 | 0.6857 | 0.7796 | 0.4 | 0.5744 | 0.8083 | 0.7386 | 0.9078 | 0.5779 | 0.8216 | 0.104 | 0.6929 | -1.0 | -1.0 | 0.2396 | 0.6681 | 0.6406 | 0.7876 | 0.1982 | 0.8 |
|
84 |
+
| 13.6203 | 7.0 | 350 | 10.0281 | 0.4716 | 0.6439 | 0.5424 | 0.32 | 0.1287 | 0.4936 | 0.52 | 0.6664 | 0.7514 | 0.4 | 0.4966 | 0.7923 | 0.7501 | 0.895 | 0.5492 | 0.8137 | 0.0383 | 0.6071 | -1.0 | -1.0 | 0.2462 | 0.6347 | 0.6563 | 0.7912 | 0.5894 | 0.7667 |
|
85 |
+
| 12.792 | 8.0 | 400 | 9.6988 | 0.4694 | 0.63 | 0.5055 | 0.0583 | 0.1478 | 0.4985 | 0.5578 | 0.7228 | 0.7949 | 0.35 | 0.5909 | 0.825 | 0.7629 | 0.9092 | 0.5704 | 0.8186 | 0.0824 | 0.7071 | -1.0 | -1.0 | 0.2587 | 0.6389 | 0.6866 | 0.7956 | 0.4551 | 0.9 |
|
86 |
+
| 12.2254 | 9.0 | 450 | 9.6264 | 0.5036 | 0.6662 | 0.5728 | 0.175 | 0.1921 | 0.5289 | 0.5642 | 0.7198 | 0.7858 | 0.35 | 0.5449 | 0.8252 | 0.7532 | 0.9007 | 0.5675 | 0.8275 | 0.0833 | 0.6714 | -1.0 | -1.0 | 0.2822 | 0.6194 | 0.6842 | 0.7956 | 0.6513 | 0.9 |
|
87 |
+
| 11.7667 | 10.0 | 500 | 9.6943 | 0.5019 | 0.6587 | 0.5594 | 0.2 | 0.1466 | 0.5289 | 0.5637 | 0.7214 | 0.7844 | 0.4 | 0.5023 | 0.826 | 0.7483 | 0.9 | 0.5688 | 0.8225 | 0.1236 | 0.6357 | -1.0 | -1.0 | 0.2696 | 0.6236 | 0.668 | 0.7912 | 0.6334 | 0.9333 |
|
88 |
+
|
89 |
+
|
90 |
+
### Framework versions
|
91 |
+
|
92 |
+
- Transformers 4.51.3
|
93 |
+
- Pytorch 2.6.0+cu124
|
94 |
+
- Datasets 3.6.0
|
95 |
+
- Tokenizers 0.21.1
|
config.json
ADDED
@@ -0,0 +1,136 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"activation_dropout": 0.0,
|
3 |
+
"activation_function": "silu",
|
4 |
+
"anchor_image_size": null,
|
5 |
+
"architectures": [
|
6 |
+
"RTDetrV2ForObjectDetection"
|
7 |
+
],
|
8 |
+
"attention_dropout": 0.0,
|
9 |
+
"auxiliary_loss": true,
|
10 |
+
"backbone": null,
|
11 |
+
"backbone_config": {
|
12 |
+
"depths": [
|
13 |
+
3,
|
14 |
+
4,
|
15 |
+
6,
|
16 |
+
3
|
17 |
+
],
|
18 |
+
"downsample_in_bottleneck": false,
|
19 |
+
"downsample_in_first_stage": false,
|
20 |
+
"embedding_size": 64,
|
21 |
+
"hidden_act": "relu",
|
22 |
+
"hidden_sizes": [
|
23 |
+
256,
|
24 |
+
512,
|
25 |
+
1024,
|
26 |
+
2048
|
27 |
+
],
|
28 |
+
"layer_type": "bottleneck",
|
29 |
+
"model_type": "rt_detr_resnet",
|
30 |
+
"num_channels": 3,
|
31 |
+
"out_features": [
|
32 |
+
"stage2",
|
33 |
+
"stage3",
|
34 |
+
"stage4"
|
35 |
+
],
|
36 |
+
"out_indices": [
|
37 |
+
2,
|
38 |
+
3,
|
39 |
+
4
|
40 |
+
],
|
41 |
+
"stage_names": [
|
42 |
+
"stem",
|
43 |
+
"stage1",
|
44 |
+
"stage2",
|
45 |
+
"stage3",
|
46 |
+
"stage4"
|
47 |
+
]
|
48 |
+
},
|
49 |
+
"backbone_kwargs": null,
|
50 |
+
"batch_norm_eps": 1e-05,
|
51 |
+
"box_noise_scale": 1.0,
|
52 |
+
"d_model": 256,
|
53 |
+
"decoder_activation_function": "relu",
|
54 |
+
"decoder_attention_heads": 8,
|
55 |
+
"decoder_ffn_dim": 1024,
|
56 |
+
"decoder_in_channels": [
|
57 |
+
256,
|
58 |
+
256,
|
59 |
+
256
|
60 |
+
],
|
61 |
+
"decoder_layers": 6,
|
62 |
+
"decoder_method": "default",
|
63 |
+
"decoder_n_levels": 3,
|
64 |
+
"decoder_n_points": 4,
|
65 |
+
"decoder_offset_scale": 0.5,
|
66 |
+
"disable_custom_kernels": true,
|
67 |
+
"dropout": 0.0,
|
68 |
+
"encode_proj_layers": [
|
69 |
+
2
|
70 |
+
],
|
71 |
+
"encoder_activation_function": "gelu",
|
72 |
+
"encoder_attention_heads": 8,
|
73 |
+
"encoder_ffn_dim": 1024,
|
74 |
+
"encoder_hidden_dim": 256,
|
75 |
+
"encoder_in_channels": [
|
76 |
+
512,
|
77 |
+
1024,
|
78 |
+
2048
|
79 |
+
],
|
80 |
+
"encoder_layers": 1,
|
81 |
+
"eos_coefficient": 0.0001,
|
82 |
+
"eval_size": null,
|
83 |
+
"feat_strides": [
|
84 |
+
8,
|
85 |
+
16,
|
86 |
+
32
|
87 |
+
],
|
88 |
+
"focal_loss_alpha": 0.75,
|
89 |
+
"focal_loss_gamma": 2.0,
|
90 |
+
"freeze_backbone_batch_norms": true,
|
91 |
+
"hidden_expansion": 1.0,
|
92 |
+
"id2label": {
|
93 |
+
"0": "bin",
|
94 |
+
"1": "hand",
|
95 |
+
"2": "not_bin",
|
96 |
+
"3": "not_hand",
|
97 |
+
"4": "not_trash",
|
98 |
+
"5": "trash",
|
99 |
+
"6": "trash_arm"
|
100 |
+
},
|
101 |
+
"initializer_bias_prior_prob": null,
|
102 |
+
"initializer_range": 0.01,
|
103 |
+
"is_encoder_decoder": true,
|
104 |
+
"label2id": {
|
105 |
+
"bin": 0,
|
106 |
+
"hand": 1,
|
107 |
+
"not_bin": 2,
|
108 |
+
"not_hand": 3,
|
109 |
+
"not_trash": 4,
|
110 |
+
"trash": 5,
|
111 |
+
"trash_arm": 6
|
112 |
+
},
|
113 |
+
"label_noise_ratio": 0.5,
|
114 |
+
"layer_norm_eps": 1e-05,
|
115 |
+
"learn_initial_query": false,
|
116 |
+
"matcher_alpha": 0.25,
|
117 |
+
"matcher_bbox_cost": 5.0,
|
118 |
+
"matcher_class_cost": 2.0,
|
119 |
+
"matcher_gamma": 2.0,
|
120 |
+
"matcher_giou_cost": 2.0,
|
121 |
+
"model_type": "rt_detr_v2",
|
122 |
+
"normalize_before": false,
|
123 |
+
"num_denoising": 100,
|
124 |
+
"num_feature_levels": 3,
|
125 |
+
"num_queries": 300,
|
126 |
+
"positional_encoding_temperature": 10000,
|
127 |
+
"torch_dtype": "float32",
|
128 |
+
"transformers_version": "4.51.3",
|
129 |
+
"use_focal_loss": true,
|
130 |
+
"use_pretrained_backbone": false,
|
131 |
+
"use_timm_backbone": false,
|
132 |
+
"weight_loss_bbox": 5.0,
|
133 |
+
"weight_loss_giou": 2.0,
|
134 |
+
"weight_loss_vfl": 1.0,
|
135 |
+
"with_box_refine": true
|
136 |
+
}
|
model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:70fa906d2fb066b8e213bd86e58a39810f1041cee76dcdf864c9e2a28bb03c04
|
3 |
+
size 171576780
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:be41a9e2dcd84709f310fa7472af851f19b314be9e5f43e4ae8da5aee22afd95
|
3 |
+
size 5368
|