Upload TableTransformerForObjectDetection

config.json

@@ -7,11 +7,12 @@
   ],
   "attention_dropout": 0.0,
   "auxiliary_loss": false,
-  "backbone": "resnet50",
+  "backbone": "resnet18",
   "backbone_config": null,
   "backbone_kwargs": null,
   "bbox_cost": 5,
   "bbox_loss_coefficient": 5,
+  "ce_loss_coefficient": 1,
   "class_cost": 1,
   "d_model": 256,
   "decoder_attention_heads": 8,
@@ -25,34 +26,36 @@
   "encoder_ffn_dim": 2048,
   "encoder_layerdrop": 0.0,
   "encoder_layers": 6,
-  "eos_coefficient": 0.1,
+  "eos_coefficient": 0.4,
   "giou_cost": 2,
   "giou_loss_coefficient": 2,
   "id2label": {
-    "1": "table",
-    "2": "table column",
-    "3": "table row",
-    "4": "table column header",
-    "5": "table projected row header",
-    "6": "table spanning cell"
+    "0": "table",
+    "1": "table column",
+    "2": "table row",
+    "3": "table column header",
+    "4": "table projected row header",
+    "5": "table spanning cell"
   },
   "init_std": 0.02,
   "init_xavier_std": 1.0,
   "is_encoder_decoder": true,
   "label2id": {
-    "table": 1,
-    "table column": 2,
-    "table column header": 4,
-    "table projected row header": 5,
-    "table row": 3,
-    "table spanning cell": 6
+    "table": 0,
+    "table column": 1,
+    "table column header": 3,
+    "table projected row header": 4,
+    "table row": 2,
+    "table spanning cell": 5
   },
   "mask_loss_coefficient": 1,
+  "max_position_embeddings": 1024,
   "model_type": "table-transformer",
   "num_channels": 3,
   "num_hidden_layers": 6,
-  "num_queries": 100,
+  "num_queries": 125,
   "position_embedding_type": "sine",
+  "scale_embedding": false,
   "torch_dtype": "float32",
   "transformers_version": "4.40.0",
   "use_pretrained_backbone": true,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2a2f190318610c1d8508f94ab04df6e2948f3d5159bff085f377d0d6cb388c6e
-size 166502212
+oid sha256:deeb81c6f6237e563bb14210c1e7c9285dfd53cbaa5261e05b7edad69dbdef14
+size 115433860