Hugging Face's logo

onnx-community
/
rtdetr_v2_r50vd-ONNX

Object Detection
Transformers.js
ONNX
rt_detr_v2
Model card Files Community
Xenova HF Staff commited on
Commit
23e2c91
·
verified ·
1 Parent(s): 486ae37

Upload folder using huggingface_hub

Browse files
Files changed (11) hide show
  1. config.json +283 -0
  2. onnx/model.onnx +3 -0
  3. onnx/model_bnb4.onnx +3 -0
  4. onnx/model_fp16.onnx +3 -0
  5. onnx/model_int8.onnx +3 -0
  6. onnx/model_q4.onnx +3 -0
  7. onnx/model_q4f16.onnx +3 -0
  8. onnx/model_quantized.onnx +3 -0
  9. onnx/model_uint8.onnx +3 -0
  10. preprocessor_config.json +26 -0
  11. quantize_config.json +18 -0
config.json ADDED
@@ -0,0 +1,283 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_attn_implementation_autoset": true,
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "silu",
5
+ "anchor_image_size": null,
6
+ "architectures": [
7
+ "RtDetrV2ForObjectDetection"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "auxiliary_loss": true,
11
+ "backbone": null,
12
+ "backbone_config": {
13
+ "depths": [
14
+ 3,
15
+ 4,
16
+ 6,
17
+ 3
18
+ ],
19
+ "downsample_in_bottleneck": false,
20
+ "downsample_in_first_stage": false,
21
+ "embedding_size": 64,
22
+ "hidden_act": "relu",
23
+ "hidden_sizes": [
24
+ 256,
25
+ 512,
26
+ 1024,
27
+ 2048
28
+ ],
29
+ "layer_type": "bottleneck",
30
+ "model_type": "rt_detr_resnet",
31
+ "num_channels": 3,
32
+ "out_features": [
33
+ "stage2",
34
+ "stage3",
35
+ "stage4"
36
+ ],
37
+ "out_indices": [
38
+ 2,
39
+ 3,
40
+ 4
41
+ ],
42
+ "stage_names": [
43
+ "stem",
44
+ "stage1",
45
+ "stage2",
46
+ "stage3",
47
+ "stage4"
48
+ ]
49
+ },
50
+ "backbone_kwargs": null,
51
+ "batch_norm_eps": 1e-05,
52
+ "box_noise_scale": 1.0,
53
+ "d_model": 256,
54
+ "decoder_activation_function": "relu",
55
+ "decoder_attention_heads": 8,
56
+ "decoder_ffn_dim": 1024,
57
+ "decoder_in_channels": [
58
+ 256,
59
+ 256,
60
+ 256
61
+ ],
62
+ "decoder_layers": 6,
63
+ "decoder_method": "default",
64
+ "decoder_n_levels": 3,
65
+ "decoder_n_points": 4,
66
+ "decoder_offset_scale": 0.5,
67
+ "disable_custom_kernels": true,
68
+ "dropout": 0.0,
69
+ "encode_proj_layers": [
70
+ 2
71
+ ],
72
+ "encoder_activation_function": "gelu",
73
+ "encoder_attention_heads": 8,
74
+ "encoder_ffn_dim": 1024,
75
+ "encoder_hidden_dim": 256,
76
+ "encoder_in_channels": [
77
+ 512,
78
+ 1024,
79
+ 2048
80
+ ],
81
+ "encoder_layers": 1,
82
+ "eos_coefficient": 0.0001,
83
+ "eval_size": null,
84
+ "feat_strides": [
85
+ 8,
86
+ 16,
87
+ 32
88
+ ],
89
+ "focal_loss_alpha": 0.75,
90
+ "focal_loss_gamma": 2.0,
91
+ "freeze_backbone_batch_norms": true,
92
+ "hidden_expansion": 1.0,
93
+ "id2label": {
94
+ "0": "person",
95
+ "1": "bicycle",
96
+ "2": "car",
97
+ "3": "motorbike",
98
+ "4": "aeroplane",
99
+ "5": "bus",
100
+ "6": "train",
101
+ "7": "truck",
102
+ "8": "boat",
103
+ "9": "traffic light",
104
+ "10": "fire hydrant",
105
+ "11": "stop sign",
106
+ "12": "parking meter",
107
+ "13": "bench",
108
+ "14": "bird",
109
+ "15": "cat",
110
+ "16": "dog",
111
+ "17": "horse",
112
+ "18": "sheep",
113
+ "19": "cow",
114
+ "20": "elephant",
115
+ "21": "bear",
116
+ "22": "zebra",
117
+ "23": "giraffe",
118
+ "24": "backpack",
119
+ "25": "umbrella",
120
+ "26": "handbag",
121
+ "27": "tie",
122
+ "28": "suitcase",
123
+ "29": "frisbee",
124
+ "30": "skis",
125
+ "31": "snowboard",
126
+ "32": "sports ball",
127
+ "33": "kite",
128
+ "34": "baseball bat",
129
+ "35": "baseball glove",
130
+ "36": "skateboard",
131
+ "37": "surfboard",
132
+ "38": "tennis racket",
133
+ "39": "bottle",
134
+ "40": "wine glass",
135
+ "41": "cup",
136
+ "42": "fork",
137
+ "43": "knife",
138
+ "44": "spoon",
139
+ "45": "bowl",
140
+ "46": "banana",
141
+ "47": "apple",
142
+ "48": "sandwich",
143
+ "49": "orange",
144
+ "50": "broccoli",
145
+ "51": "carrot",
146
+ "52": "hot dog",
147
+ "53": "pizza",
148
+ "54": "donut",
149
+ "55": "cake",
150
+ "56": "chair",
151
+ "57": "sofa",
152
+ "58": "pottedplant",
153
+ "59": "bed",
154
+ "60": "diningtable",
155
+ "61": "toilet",
156
+ "62": "tvmonitor",
157
+ "63": "laptop",
158
+ "64": "mouse",
159
+ "65": "remote",
160
+ "66": "keyboard",
161
+ "67": "cell phone",
162
+ "68": "microwave",
163
+ "69": "oven",
164
+ "70": "toaster",
165
+ "71": "sink",
166
+ "72": "refrigerator",
167
+ "73": "book",
168
+ "74": "clock",
169
+ "75": "vase",
170
+ "76": "scissors",
171
+ "77": "teddy bear",
172
+ "78": "hair drier",
173
+ "79": "toothbrush"
174
+ },
175
+ "initializer_bias_prior_prob": null,
176
+ "initializer_range": 0.01,
177
+ "is_encoder_decoder": true,
178
+ "label2id": {
179
+ "aeroplane": 4,
180
+ "apple": 47,
181
+ "backpack": 24,
182
+ "banana": 46,
183
+ "baseball bat": 34,
184
+ "baseball glove": 35,
185
+ "bear": 21,
186
+ "bed": 59,
187
+ "bench": 13,
188
+ "bicycle": 1,
189
+ "bird": 14,
190
+ "boat": 8,
191
+ "book": 73,
192
+ "bottle": 39,
193
+ "bowl": 45,
194
+ "broccoli": 50,
195
+ "bus": 5,
196
+ "cake": 55,
197
+ "car": 2,
198
+ "carrot": 51,
199
+ "cat": 15,
200
+ "cell phone": 67,
201
+ "chair": 56,
202
+ "clock": 74,
203
+ "cow": 19,
204
+ "cup": 41,
205
+ "diningtable": 60,
206
+ "dog": 16,
207
+ "donut": 54,
208
+ "elephant": 20,
209
+ "fire hydrant": 10,
210
+ "fork": 42,
211
+ "frisbee": 29,
212
+ "giraffe": 23,
213
+ "hair drier": 78,
214
+ "handbag": 26,
215
+ "horse": 17,
216
+ "hot dog": 52,
217
+ "keyboard": 66,
218
+ "kite": 33,
219
+ "knife": 43,
220
+ "laptop": 63,
221
+ "microwave": 68,
222
+ "motorbike": 3,
223
+ "mouse": 64,
224
+ "orange": 49,
225
+ "oven": 69,
226
+ "parking meter": 12,
227
+ "person": 0,
228
+ "pizza": 53,
229
+ "pottedplant": 58,
230
+ "refrigerator": 72,
231
+ "remote": 65,
232
+ "sandwich": 48,
233
+ "scissors": 76,
234
+ "sheep": 18,
235
+ "sink": 71,
236
+ "skateboard": 36,
237
+ "skis": 30,
238
+ "snowboard": 31,
239
+ "sofa": 57,
240
+ "spoon": 44,
241
+ "sports ball": 32,
242
+ "stop sign": 11,
243
+ "suitcase": 28,
244
+ "surfboard": 37,
245
+ "teddy bear": 77,
246
+ "tennis racket": 38,
247
+ "tie": 27,
248
+ "toaster": 70,
249
+ "toilet": 61,
250
+ "toothbrush": 79,
251
+ "traffic light": 9,
252
+ "train": 6,
253
+ "truck": 7,
254
+ "tvmonitor": 62,
255
+ "umbrella": 25,
256
+ "vase": 75,
257
+ "wine glass": 40,
258
+ "zebra": 22
259
+ },
260
+ "label_noise_ratio": 0.5,
261
+ "layer_norm_eps": 1e-05,
262
+ "learn_initial_query": false,
263
+ "matcher_alpha": 0.25,
264
+ "matcher_bbox_cost": 5.0,
265
+ "matcher_class_cost": 2.0,
266
+ "matcher_gamma": 2.0,
267
+ "matcher_giou_cost": 2.0,
268
+ "model_type": "rt_detr_v2",
269
+ "normalize_before": false,
270
+ "num_denoising": 100,
271
+ "num_feature_levels": 3,
272
+ "num_queries": 300,
273
+ "positional_encoding_temperature": 10000,
274
+ "torch_dtype": "float32",
275
+ "transformers_version": "4.50.3",
276
+ "use_focal_loss": true,
277
+ "use_pretrained_backbone": false,
278
+ "use_timm_backbone": false,
279
+ "weight_loss_bbox": 5.0,
280
+ "weight_loss_giou": 2.0,
281
+ "weight_loss_vfl": 1.0,
282
+ "with_box_refine": true
283
+ }
onnx/model.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d6d5ce9f633189539bc892c70b6a60120a32a9799798811daeb8456f9b9e2c15
3
+ size 172305665
onnx/model_bnb4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5b0f56213951072ee71159bcf8387590ca378e87e74519d1ccdad5bc7c03b9cf
3
+ size 144820158
onnx/model_fp16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f94f4d84e9c7258e408af315c20395a0e835e324c1e4aa7cd23bd5f93258b16
3
+ size 86484200
onnx/model_int8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:be2127de397924baf205f968befe28a265598fd52ea3de95a5502617731fe0a3
3
+ size 44326915
onnx/model_q4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d6e8e2ca01d39acbbe0a85a62a29259d8c28e1cf42cf258efd77552cff1eac1a
3
+ size 145329534
onnx/model_q4f16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:efa3fdf3991e6493c6f0a934ff70a7bf3be0320c9bffd6e086caa82f9c2984ab
3
+ size 75008149
onnx/model_quantized.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:33276b82e0f858f5a6d00996afde5508a51e6a080cfda116eff722776bbbdabf
3
+ size 44326915
onnx/model_uint8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:33276b82e0f858f5a6d00996afde5508a51e6a080cfda116eff722776bbbdabf
3
+ size 44326915
preprocessor_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_convert_annotations": true,
3
+ "do_normalize": false,
4
+ "do_pad": false,
5
+ "do_rescale": true,
6
+ "do_resize": true,
7
+ "format": "coco_detection",
8
+ "image_mean": [
9
+ 0.485,
10
+ 0.456,
11
+ 0.406
12
+ ],
13
+ "image_processor_type": "RTDetrImageProcessor",
14
+ "image_std": [
15
+ 0.229,
16
+ 0.224,
17
+ 0.225
18
+ ],
19
+ "pad_size": null,
20
+ "resample": 2,
21
+ "rescale_factor": 0.00392156862745098,
22
+ "size": {
23
+ "height": 640,
24
+ "width": 640
25
+ }
26
+ }
quantize_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "modes": [
3
+ "fp16",
4
+ "q8",
5
+ "int8",
6
+ "uint8",
7
+ "q4",
8
+ "q4f16",
9
+ "bnb4"
10
+ ],
11
+ "per_channel": true,
12
+ "reduce_range": true,
13
+ "block_size": null,
14
+ "is_symmetric": true,
15
+ "accuracy_level": null,
16
+ "quant_type": 1,
17
+ "op_block_list": null
18
+ }