{ "_name_or_path": "facebook/detr-resnet-50", "activation_dropout": 0.0, "activation_function": "relu", "architectures": [ "DetrForObjectDetection" ], "attention_dropout": 0.0, "auxiliary_loss": false, "backbone": "resnet50", "backbone_config": null, "backbone_kwargs": { "in_chans": 3, "out_indices": [ 1, 2, 3, 4 ] }, "bbox_cost": 5, "bbox_loss_coefficient": 5, "class_cost": 1, "classifier_dropout": 0.0, "d_model": 256, "decoder_attention_heads": 8, "decoder_ffn_dim": 2048, "decoder_layerdrop": 0.0, "decoder_layers": 6, "dice_loss_coefficient": 1, "dilation": false, "dropout": 0.1, "encoder_attention_heads": 8, "encoder_ffn_dim": 2048, "encoder_layerdrop": 0.0, "encoder_layers": 6, "eos_coefficient": 0.1, "giou_cost": 2, "giou_loss_coefficient": 2, "id2label": { "1": 10, "2": 50, "3": 39, "4": 40, "5": 22, "6": 20, "7": 19, "8": 26, "9": 27, "10": 31, "11": 28, "12": 30, "13": 69, "14": 68, "15": 15, "16": 12, "17": 8, "18": 4, "19": 32, "20": 33, "21": 3, "22": 63, "23": 64, "24": 87, "25": 89, "26": 57, "27": 58, "28": 83, "29": 23, "30": 24, "31": 61, "32": 62, "33": 101, "34": 100, "35": 102, "36": 63, "37": 66, "38": 25, "39": 23, "40": 24, "41": 51, "42": 14, "43": 84, "44": 92, "45": 47, "46": 48, "47": 54, "48": 55, "49": 56, "50": 20, "51": 19, "52": 28, "53": 29, "54": 83, "55": 18, "56": 53, "57": 81, "58": 80, "59": 78, "60": 8, "61": 52, "62": 42, "63": 43, "64": 53, "65": 61, "66": 62, "67": 106, "68": 107, "69": 2, "70": 74, "71": 75, "72": 32, "73": 35, "74": 34, "75": 20, "76": 19, "77": 59, "78": 60, "79": 97, "80": 98, "81": 99, "82": 57, "83": 58, "84": 6, "85": 101, "86": 100, "87": 102, "88": 67, "89": 26, "90": 97, "91": 98, "92": 99, "93": 92, "94": 93, "95": 1, "96": 53, "97": 98, "98": 99, "99": 95, "100": 96, "101": 21, "102": 72, "103": 10, "104": 76, "105": 77, "106": 90, "107": 91, "108": 7, "109": 72, "110": 73, "111": 14, "112": 16, "113": 87, "114": 67, "115": 85, "116": 86, "117": 82, "118": 70, "119": 71, "120": 41, "121": 42, "122": 49, "123": 74, "124": 63, "125": 63, "126": 81, "127": 80, "128": 78, "129": 79, "130": 35, "131": 34, "132": 106, "133": 107, "134": 2, "135": 15, "136": 12, "137": 36, "138": 37, "139": 38, "140": 28, "141": 29, "142": 103, "143": 104, "144": 105, "145": 74, "146": 51, "147": 70, "148": 87, "149": 88, "150": 7, "151": 9, "152": 94, "153": 92, "154": 11, "155": 13, "156": 17, "157": 44, "158": 45, "159": 46 }, "init_std": 0.02, "init_xavier_std": 1.0, "is_encoder_decoder": true, "label2id": { "1": 95, "2": 134, "3": 21, "4": 18, "6": 84, "7": 150, "8": 60, "9": 151, "10": 103, "11": 154, "12": 136, "13": 155, "14": 111, "15": 135, "16": 112, "17": 156, "18": 55, "19": 76, "20": 75, "21": 101, "22": 5, "23": 39, "24": 40, "25": 38, "26": 89, "27": 9, "28": 140, "29": 141, "30": 12, "31": 10, "32": 72, "33": 20, "34": 131, "35": 130, "36": 137, "37": 138, "38": 139, "39": 3, "40": 4, "41": 120, "42": 121, "43": 63, "44": 157, "45": 158, "46": 159, "47": 45, "48": 46, "49": 122, "50": 2, "51": 146, "52": 61, "53": 96, "54": 47, "55": 48, "56": 49, "57": 82, "58": 83, "59": 77, "60": 78, "61": 65, "62": 66, "63": 125, "64": 23, "66": 37, "67": 114, "68": 14, "69": 13, "70": 147, "71": 119, "72": 109, "73": 110, "74": 145, "75": 71, "76": 104, "77": 105, "78": 128, "79": 129, "80": 127, "81": 126, "82": 117, "83": 54, "84": 43, "85": 115, "86": 116, "87": 148, "88": 149, "89": 25, "90": 106, "91": 107, "92": 153, "93": 94, "94": 152, "95": 99, "96": 100, "97": 90, "98": 97, "99": 98, "100": 86, "101": 85, "102": 87, "103": 142, "104": 143, "105": 144, "106": 132, "107": 133 }, "mask_loss_coefficient": 1, "max_position_embeddings": 1024, "model_type": "detr", "num_channels": 3, "num_hidden_layers": 6, "num_queries": 100, "position_embedding_type": "sine", "scale_embedding": false, "torch_dtype": "float32", "transformers_version": "4.41.1", "use_pretrained_backbone": true, "use_timm_backbone": true }