Hugging Face's logo

onnx-community
/
rtdetr_r18vd

Object Detection
Transformers.js
ONNX
rt_detr
Model card Files Community
1
Xenova HF staff commited on
Commit
ec641af
•
1 Parent(s): 2acdadb

Upload folder using huggingface_hub

Browse files
Files changed (10) hide show
  1. config.json +268 -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_quantized.onnx +3 -0
  8. onnx/model_uint8.onnx +3 -0
  9. preprocessor_config.json +26 -0
  10. quantize_config.json +141 -0
config.json ADDED
@@ -0,0 +1,268 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "PekingU/rtdetr_r18vd",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "silu",
5
+ "anchor_image_size": [
6
+ 640,
7
+ 640
8
+ ],
9
+ "architectures": [
10
+ "RTDetrForObjectDetection"
11
+ ],
12
+ "attention_dropout": 0.0,
13
+ "auxiliary_loss": true,
14
+ "backbone": null,
15
+ "backbone_config": {
16
+ "depths": [
17
+ 2,
18
+ 2,
19
+ 2,
20
+ 2
21
+ ],
22
+ "hidden_sizes": [
23
+ 64,
24
+ 128,
25
+ 256,
26
+ 512
27
+ ],
28
+ "layer_type": "basic",
29
+ "model_type": "rt_detr_resnet",
30
+ "out_features": [
31
+ "stage2",
32
+ "stage3",
33
+ "stage4"
34
+ ],
35
+ "out_indices": [
36
+ 2,
37
+ 3,
38
+ 4
39
+ ]
40
+ },
41
+ "backbone_kwargs": null,
42
+ "batch_norm_eps": 1e-05,
43
+ "box_noise_scale": 1.0,
44
+ "d_model": 256,
45
+ "decoder_activation_function": "relu",
46
+ "decoder_attention_heads": 8,
47
+ "decoder_ffn_dim": 1024,
48
+ "decoder_in_channels": [
49
+ 256,
50
+ 256,
51
+ 256
52
+ ],
53
+ "decoder_layers": 3,
54
+ "decoder_n_points": 4,
55
+ "disable_custom_kernels": true,
56
+ "dropout": 0.0,
57
+ "encode_proj_layers": [
58
+ 2
59
+ ],
60
+ "encoder_activation_function": "gelu",
61
+ "encoder_attention_heads": 8,
62
+ "encoder_ffn_dim": 1024,
63
+ "encoder_hidden_dim": 256,
64
+ "encoder_in_channels": [
65
+ 128,
66
+ 256,
67
+ 512
68
+ ],
69
+ "encoder_layers": 1,
70
+ "eos_coefficient": 0.0001,
71
+ "eval_size": null,
72
+ "feat_strides": [
73
+ 8,
74
+ 16,
75
+ 32
76
+ ],
77
+ "focal_loss_alpha": 0.75,
78
+ "focal_loss_gamma": 2.0,
79
+ "hidden_expansion": 0.5,
80
+ "id2label": {
81
+ "0": "person",
82
+ "1": "bicycle",
83
+ "2": "car",
84
+ "3": "motorbike",
85
+ "4": "aeroplane",
86
+ "5": "bus",
87
+ "6": "train",
88
+ "7": "truck",
89
+ "8": "boat",
90
+ "9": "traffic light",
91
+ "10": "fire hydrant",
92
+ "11": "stop sign",
93
+ "12": "parking meter",
94
+ "13": "bench",
95
+ "14": "bird",
96
+ "15": "cat",
97
+ "16": "dog",
98
+ "17": "horse",
99
+ "18": "sheep",
100
+ "19": "cow",
101
+ "20": "elephant",
102
+ "21": "bear",
103
+ "22": "zebra",
104
+ "23": "giraffe",
105
+ "24": "backpack",
106
+ "25": "umbrella",
107
+ "26": "handbag",
108
+ "27": "tie",
109
+ "28": "suitcase",
110
+ "29": "frisbee",
111
+ "30": "skis",
112
+ "31": "snowboard",
113
+ "32": "sports ball",
114
+ "33": "kite",
115
+ "34": "baseball bat",
116
+ "35": "baseball glove",
117
+ "36": "skateboard",
118
+ "37": "surfboard",
119
+ "38": "tennis racket",
120
+ "39": "bottle",
121
+ "40": "wine glass",
122
+ "41": "cup",
123
+ "42": "fork",
124
+ "43": "knife",
125
+ "44": "spoon",
126
+ "45": "bowl",
127
+ "46": "banana",
128
+ "47": "apple",
129
+ "48": "sandwich",
130
+ "49": "orange",
131
+ "50": "broccoli",
132
+ "51": "carrot",
133
+ "52": "hot dog",
134
+ "53": "pizza",
135
+ "54": "donut",
136
+ "55": "cake",
137
+ "56": "chair",
138
+ "57": "sofa",
139
+ "58": "pottedplant",
140
+ "59": "bed",
141
+ "60": "diningtable",
142
+ "61": "toilet",
143
+ "62": "tvmonitor",
144
+ "63": "laptop",
145
+ "64": "mouse",
146
+ "65": "remote",
147
+ "66": "keyboard",
148
+ "67": "cell phone",
149
+ "68": "microwave",
150
+ "69": "oven",
151
+ "70": "toaster",
152
+ "71": "sink",
153
+ "72": "refrigerator",
154
+ "73": "book",
155
+ "74": "clock",
156
+ "75": "vase",
157
+ "76": "scissors",
158
+ "77": "teddy bear",
159
+ "78": "hair drier",
160
+ "79": "toothbrush"
161
+ },
162
+ "initializer_range": 0.01,
163
+ "is_encoder_decoder": true,
164
+ "label2id": {
165
+ "aeroplane": 4,
166
+ "apple": 47,
167
+ "backpack": 24,
168
+ "banana": 46,
169
+ "baseball bat": 34,
170
+ "baseball glove": 35,
171
+ "bear": 21,
172
+ "bed": 59,
173
+ "bench": 13,
174
+ "bicycle": 1,
175
+ "bird": 14,
176
+ "boat": 8,
177
+ "book": 73,
178
+ "bottle": 39,
179
+ "bowl": 45,
180
+ "broccoli": 50,
181
+ "bus": 5,
182
+ "cake": 55,
183
+ "car": 2,
184
+ "carrot": 51,
185
+ "cat": 15,
186
+ "cell phone": 67,
187
+ "chair": 56,
188
+ "clock": 74,
189
+ "cow": 19,
190
+ "cup": 41,
191
+ "diningtable": 60,
192
+ "dog": 16,
193
+ "donut": 54,
194
+ "elephant": 20,
195
+ "fire hydrant": 10,
196
+ "fork": 42,
197
+ "frisbee": 29,
198
+ "giraffe": 23,
199
+ "hair drier": 78,
200
+ "handbag": 26,
201
+ "horse": 17,
202
+ "hot dog": 52,
203
+ "keyboard": 66,
204
+ "kite": 33,
205
+ "knife": 43,
206
+ "laptop": 63,
207
+ "microwave": 68,
208
+ "motorbike": 3,
209
+ "mouse": 64,
210
+ "orange": 49,
211
+ "oven": 69,
212
+ "parking meter": 12,
213
+ "person": 0,
214
+ "pizza": 53,
215
+ "pottedplant": 58,
216
+ "refrigerator": 72,
217
+ "remote": 65,
218
+ "sandwich": 48,
219
+ "scissors": 76,
220
+ "sheep": 18,
221
+ "sink": 71,
222
+ "skateboard": 36,
223
+ "skis": 30,
224
+ "snowboard": 31,
225
+ "sofa": 57,
226
+ "spoon": 44,
227
+ "sports ball": 32,
228
+ "stop sign": 11,
229
+ "suitcase": 28,
230
+ "surfboard": 37,
231
+ "teddy bear": 77,
232
+ "tennis racket": 38,
233
+ "tie": 27,
234
+ "toaster": 70,
235
+ "toilet": 61,
236
+ "toothbrush": 79,
237
+ "traffic light": 9,
238
+ "train": 6,
239
+ "truck": 7,
240
+ "tvmonitor": 62,
241
+ "umbrella": 25,
242
+ "vase": 75,
243
+ "wine glass": 40,
244
+ "zebra": 22
245
+ },
246
+ "label_noise_ratio": 0.5,
247
+ "layer_norm_eps": 1e-05,
248
+ "learn_initial_query": false,
249
+ "matcher_alpha": 0.25,
250
+ "matcher_bbox_cost": 5.0,
251
+ "matcher_class_cost": 2.0,
252
+ "matcher_gamma": 2.0,
253
+ "matcher_giou_cost": 2.0,
254
+ "model_type": "rt_detr",
255
+ "normalize_before": false,
256
+ "num_denoising": 100,
257
+ "num_feature_levels": 3,
258
+ "num_queries": 300,
259
+ "positional_encoding_temperature": 10000,
260
+ "transformers_version": "4.42.2",
261
+ "use_focal_loss": true,
262
+ "use_pretrained_backbone": false,
263
+ "use_timm_backbone": false,
264
+ "weight_loss_bbox": 5.0,
265
+ "weight_loss_giou": 2.0,
266
+ "weight_loss_vfl": 1.0,
267
+ "with_box_refine": true
268
+ }
onnx/model.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11843b02455cc24009aed24d4c40db721b1093be5ccd6bbe7b9c441abb1d0558
3
+ size 82572357
onnx/model_bnb4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7a14ea64534af80b31ee93566f98290680daaf268c3b54317bf2eb2b233209a7
3
+ size 65960054
onnx/model_fp16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:361ce1055638d28c3d33f4036068cd94e84baaf07d22cb4e02c6f7b5167eaf94
3
+ size 41441171
onnx/model_int8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:328dfaa2505a376b2316a8fec5dcbc52125c477d8cd744977db5922c1b4d4486
3
+ size 21713196
onnx/model_q4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4e57f08a66311788c1ebd02cb1dd86984aa15099df9d799fabbf28facef7903f
3
+ size 66288690
onnx/model_quantized.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:85703b0f56dbaceb89b21122e580fd11e11a879111fd727d0e9abdaf0e3620bf
3
+ size 21713196
onnx/model_uint8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:85703b0f56dbaceb89b21122e580fd11e11a879111fd727d0e9abdaf0e3620bf
3
+ size 21713196
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 +1,141 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "fp16": {},
3
+ "q8": {
4
+ "per_model_config": {
5
+ "model": {
6
+ "op_types": [
7
+ "Add",
8
+ "AveragePool",
9
+ "Clip",
10
+ "Concat",
11
+ "Conv",
12
+ "Div",
13
+ "Erf",
14
+ "Gather",
15
+ "GatherElements",
16
+ "GridSample",
17
+ "Log",
18
+ "MatMul",
19
+ "MaxPool",
20
+ "Mul",
21
+ "Pow",
22
+ "ReduceMax",
23
+ "ReduceMean",
24
+ "ReduceSum",
25
+ "Relu",
26
+ "Reshape",
27
+ "Resize",
28
+ "Shape",
29
+ "Sigmoid",
30
+ "Slice",
31
+ "Softmax",
32
+ "Split",
33
+ "Sqrt",
34
+ "Sub",
35
+ "Tile",
36
+ "TopK",
37
+ "Transpose",
38
+ "Unsqueeze"
39
+ ],
40
+ "weight_type": "QUInt8"
41
+ }
42
+ },
43
+ "per_channel": true,
44
+ "reduce_range": true
45
+ },
46
+ "int8": {
47
+ "per_model_config": {
48
+ "model": {
49
+ "op_types": [
50
+ "Add",
51
+ "AveragePool",
52
+ "Clip",
53
+ "Concat",
54
+ "Conv",
55
+ "Div",
56
+ "Erf",
57
+ "Gather",
58
+ "GatherElements",
59
+ "GridSample",
60
+ "Log",
61
+ "MatMul",
62
+ "MaxPool",
63
+ "Mul",
64
+ "Pow",
65
+ "ReduceMax",
66
+ "ReduceMean",
67
+ "ReduceSum",
68
+ "Relu",
69
+ "Reshape",
70
+ "Resize",
71
+ "Shape",
72
+ "Sigmoid",
73
+ "Slice",
74
+ "Softmax",
75
+ "Split",
76
+ "Sqrt",
77
+ "Sub",
78
+ "Tile",
79
+ "TopK",
80
+ "Transpose",
81
+ "Unsqueeze"
82
+ ],
83
+ "weight_type": "QInt8"
84
+ }
85
+ },
86
+ "per_channel": true,
87
+ "reduce_range": true
88
+ },
89
+ "uint8": {
90
+ "per_model_config": {
91
+ "model": {
92
+ "op_types": [
93
+ "Add",
94
+ "AveragePool",
95
+ "Clip",
96
+ "Concat",
97
+ "Conv",
98
+ "Div",
99
+ "Erf",
100
+ "Gather",
101
+ "GatherElements",
102
+ "GridSample",
103
+ "Log",
104
+ "MatMul",
105
+ "MaxPool",
106
+ "Mul",
107
+ "Pow",
108
+ "ReduceMax",
109
+ "ReduceMean",
110
+ "ReduceSum",
111
+ "Relu",
112
+ "Reshape",
113
+ "Resize",
114
+ "Shape",
115
+ "Sigmoid",
116
+ "Slice",
117
+ "Softmax",
118
+ "Split",
119
+ "Sqrt",
120
+ "Sub",
121
+ "Tile",
122
+ "TopK",
123
+ "Transpose",
124
+ "Unsqueeze"
125
+ ],
126
+ "weight_type": "QUInt8"
127
+ }
128
+ },
129
+ "per_channel": true,
130
+ "reduce_range": true
131
+ },
132
+ "q4": {
133
+ "block_size": 32,
134
+ "is_symmetric": true,
135
+ "accuracy_level": null
136
+ },
137
+ "bnb4": {
138
+ "block_size": 64,
139
+ "quant_type": 1
140
+ }
141
+ }