{ "_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": 82, "2": 94, "3": 92, "4": 53, "5": 35, "6": 34, "7": 31, "8": 11, "9": 13, "10": 63, "11": 70, "12": 71, "13": 51, "14": 95, "15": 96, "16": 10, "17": 97, "18": 98, "19": 99, "20": 101, "21": 100, "22": 102, "23": 76, "24": 77, "25": 74, "26": 18, "27": 61, "28": 62, "29": 7, "30": 9, "31": 106, "32": 107, "33": 2, "34": 44, "35": 45, "36": 46, "37": 4, "38": 57, "39": 58, "40": 35, "41": 34, "42": 42, "43": 43, "44": 51, "45": 97, "46": 98, "47": 99, "48": 85, "49": 86, "50": 20, "51": 19, "52": 59, "53": 60, "54": 47, "55": 48, "56": 14, "57": 103, "58": 104, "59": 105, "60": 63, "61": 64, "62": 67, "63": 92, "64": 15, "65": 12, "66": 74, "67": 75, "68": 22, "69": 20, "70": 19, "71": 7, "72": 74, "73": 28, "74": 29, "75": 17, "76": 23, "77": 24, "78": 66, "79": 69, "80": 52, "81": 41, "82": 42, "83": 6, "84": 54, "85": 55, "86": 56, "87": 8, "88": 81, "89": 80, "90": 78, "91": 79, "92": 57, "93": 58, "94": 106, "95": 107, "96": 2, "97": 50, "98": 49, "99": 36, "100": 37, "101": 38, "102": 1, "103": 98, "104": 99, "105": 10, "106": 28, "107": 30, "108": 68, "109": 28, "110": 29, "111": 8, "112": 87, "113": 87, "114": 88, "115": 25, "116": 23, "117": 24, "118": 81, "119": 80, "120": 78, "121": 3, "122": 32, "123": 21, "124": 15, "125": 12, "126": 87, "127": 89, "128": 63, "129": 63, "130": 72, "131": 73, "132": 26, "133": 20, "134": 19, "135": 14, "136": 16, "137": 39, "138": 40, "139": 70, "140": 83, "141": 90, "142": 91, "143": 67, "144": 53, "145": 92, "146": 93, "147": 32, "148": 33, "149": 101, "150": 100, "151": 102, "152": 53, "153": 61, "154": 62, "155": 84, "156": 83, "157": 26, "158": 27, "159": 72 }, "init_std": 0.02, "init_xavier_std": 1.0, "is_encoder_decoder": true, "label2id": { "1": 102, "2": 96, "3": 121, "4": 37, "6": 83, "7": 71, "8": 111, "9": 30, "10": 105, "11": 8, "12": 125, "13": 9, "14": 135, "15": 124, "16": 136, "17": 75, "18": 26, "19": 134, "20": 133, "21": 123, "22": 68, "23": 116, "24": 117, "25": 115, "26": 157, "27": 158, "28": 109, "29": 110, "30": 107, "31": 7, "32": 147, "33": 148, "34": 41, "35": 40, "36": 99, "37": 100, "38": 101, "39": 137, "40": 138, "41": 81, "42": 82, "43": 43, "44": 34, "45": 35, "46": 36, "47": 54, "48": 55, "49": 98, "50": 97, "51": 44, "52": 80, "53": 152, "54": 84, "55": 85, "56": 86, "57": 92, "58": 93, "59": 52, "60": 53, "61": 153, "62": 154, "63": 129, "64": 61, "66": 78, "67": 143, "68": 108, "69": 79, "70": 139, "71": 12, "72": 159, "73": 131, "74": 72, "75": 67, "76": 23, "77": 24, "78": 120, "79": 91, "80": 119, "81": 118, "82": 1, "83": 156, "84": 155, "85": 48, "86": 49, "87": 126, "88": 114, "89": 127, "90": 141, "91": 142, "92": 145, "93": 146, "94": 2, "95": 14, "96": 15, "97": 45, "98": 103, "99": 104, "100": 150, "101": 149, "102": 151, "103": 57, "104": 58, "105": 59, "106": 94, "107": 95 }, "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": "4.41.0", "use_pretrained_backbone": true, "use_timm_backbone": true }