{ "_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": 81, "2": 80, "3": 78, "4": 26, "5": 27, "6": 67, "7": 106, "8": 107, "9": 2, "10": 14, "11": 63, "12": 64, "13": 1, "14": 63, "15": 10, "16": 17, "17": 15, "18": 12, "19": 101, "20": 100, "21": 102, "22": 52, "23": 15, "24": 12, "25": 66, "26": 63, "27": 63, "28": 54, "29": 55, "30": 56, "31": 87, "32": 26, "33": 87, "34": 88, "35": 90, "36": 91, "37": 103, "38": 104, "39": 105, "40": 92, "41": 51, "42": 22, "43": 20, "44": 19, "45": 57, "46": 58, "47": 59, "48": 60, "49": 98, "50": 99, "51": 7, "52": 9, "53": 10, "54": 41, "55": 42, "56": 32, "57": 33, "58": 28, "59": 30, "60": 35, "61": 34, "62": 8, "63": 14, "64": 16, "65": 51, "66": 74, "67": 72, "68": 73, "69": 61, "70": 62, "71": 53, "72": 28, "73": 29, "74": 39, "75": 40, "76": 106, "77": 107, "78": 2, "79": 83, "80": 70, "81": 69, "82": 87, "83": 89, "84": 49, "85": 70, "86": 71, "87": 76, "88": 77, "89": 72, "90": 74, "91": 75, "92": 101, "93": 100, "94": 102, "95": 6, "96": 35, "97": 34, "98": 83, "99": 21, "100": 25, "101": 23, "102": 24, "103": 53, "104": 31, "105": 18, "106": 11, "107": 13, "108": 20, "109": 19, "110": 84, "111": 92, "112": 93, "113": 20, "114": 19, "115": 36, "116": 37, "117": 38, "118": 57, "119": 58, "120": 44, "121": 45, "122": 46, "123": 95, "124": 96, "125": 23, "126": 24, "127": 61, "128": 62, "129": 4, "130": 67, "131": 74, "132": 7, "133": 97, "134": 98, "135": 99, "136": 47, "137": 48, "138": 32, "139": 53, "140": 94, "141": 92, "142": 85, "143": 86, "144": 8, "145": 68, "146": 3, "147": 82, "148": 81, "149": 80, "150": 78, "151": 79, "152": 97, "153": 98, "154": 99, "155": 50, "156": 28, "157": 29, "158": 42, "159": 43 }, "init_std": 0.02, "init_xavier_std": 1.0, "is_encoder_decoder": true, "label2id": { "1": 13, "2": 78, "3": 146, "4": 129, "6": 95, "7": 132, "8": 144, "9": 52, "10": 53, "11": 106, "12": 24, "13": 107, "14": 63, "15": 23, "16": 64, "17": 16, "18": 105, "19": 114, "20": 113, "21": 99, "22": 42, "23": 125, "24": 126, "25": 100, "26": 32, "27": 5, "28": 156, "29": 157, "30": 59, "31": 104, "32": 138, "33": 57, "34": 97, "35": 96, "36": 115, "37": 116, "38": 117, "39": 74, "40": 75, "41": 54, "42": 158, "43": 159, "44": 120, "45": 121, "46": 122, "47": 136, "48": 137, "49": 84, "50": 155, "51": 65, "52": 22, "53": 139, "54": 28, "55": 29, "56": 30, "57": 118, "58": 119, "59": 47, "60": 48, "61": 127, "62": 128, "63": 27, "64": 12, "66": 25, "67": 130, "68": 145, "69": 81, "70": 85, "71": 86, "72": 89, "73": 68, "74": 131, "75": 91, "76": 87, "77": 88, "78": 150, "79": 151, "80": 149, "81": 148, "82": 147, "83": 98, "84": 110, "85": 142, "86": 143, "87": 82, "88": 34, "89": 83, "90": 35, "91": 36, "92": 141, "93": 112, "94": 140, "95": 123, "96": 124, "97": 152, "98": 153, "99": 154, "100": 93, "101": 92, "102": 94, "103": 37, "104": 38, "105": 39, "106": 76, "107": 77 }, "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 }