config.json

{
  "activation_dropout": 0.0,
  "activation_function": "relu",
  "architectures": [
    "GroundingDINOForObjectDetection"
  ],
  "attention_dropout": 0.0,
  "auxiliary_loss": false,
  "backbone": "swin",
  "backbone_config": {
    "depths": [
      2,
      2,
      18,
      2
    ],
    "embed_dim": 128,
    "image_size": 384,
    "model_type": "swin",
    "num_heads": [
      4,
      8,
      16,
      32
    ],
    "out_features": [
      "stage2",
      "stage3",
      "stage4"
    ],
    "out_indices": [
      2,
      3,
      4
    ],
    "window_size": 12
  },
  "bbox_cost": 5,
  "bbox_loss_coefficient": 5,
  "class_cost": 1,
  "d_model": 256,
  "decoder_attention_heads": 8,
  "decoder_bbox_embed_share": true,
  "decoder_ffn_dim": 2048,
  "decoder_layers": 6,
  "decoder_n_points": 4,
  "dice_loss_coefficient": 1,
  "dilation": false,
  "disable_custom_kernels": false,
  "dropout": 0.1,
  "embedding_init_target": true,
  "encoder_attention_heads": 8,
  "encoder_ffn_dim": 2048,
  "encoder_layerdrop": 0.0,
  "encoder_layers": 6,
  "encoder_n_points": 4,
  "eos_coefficient": 0.1,
  "focal_alpha": 0.25,
  "fusion_dropout": 0.0,
  "fusion_droppath": 0.1,
  "giou_cost": 2,
  "giou_loss_coefficient": 2,
  "init_std": 0.02,
  "init_xavier_std": 1.0,
  "is_encoder_decoder": true,
  "mask_loss_coefficient": 1,
  "max_position_embeddings": 1024,
  "max_text_len": 256,
  "model_type": "grounding-dino",
  "num_channels": 3,
  "num_feature_levels": 4,
  "num_queries": 900,
  "position_embedding_type": "sine",
  "positional_embedding_temperature": 20,
  "query_dim": 4,
  "sub_sentence_present": true,
  "text_backbone_config": {
    "_name_or_path": "bert-base-uncased",
    "architectures": [
      "BertForMaskedLM"
    ],
    "gradient_checkpointing": false,
    "model_type": "bert"
  },
  "text_enhancer_dropout": 0.0,
  "torch_dtype": "float32",
  "transformers_version": "4.33.0.dev0",
  "two_stage": true,
  "two_stage_bbox_embed_share": false,
  "two_stage_class_embed_share": false,
  "two_stage_num_proposals": 900,
  "use_pretrained_backbone": true,
  "use_timm_backbone": false,
  "with_box_refine": true
}