Fix running with --deploy --deploy-meta-path

detection/data/__init__.py

@@ -18,4 +18,9 @@ def get_dataset(split="train"):
 
     # meta['image_path'] = [f'data/image/{panoid}_{heading}.jpg' for panoid, heading in zip(meta['panoid'], meta['heading'])]
 
-    return BaseDataset(info, meta)[split]
+    ds = BaseDataset(info, meta)
+    if split:
+        return ds[split]
+    else:
+        return ds
+

detection/data/base.py

@@ -28,6 +28,8 @@ class BaseDataset(Dataset,
         if not _is_path(file_path):
             return None
         image_pil = Image.open(file_path).convert('RGB')
+        # scale larger images to fit
+        # image_pil.thumnail((640,640), Image.Resampling.LANCZOS)
         image_np = np.array(image_pil)
         return image_np
 

detection/lightning/detection.py

@@ -64,6 +64,7 @@ class DetectionTask(pl.LightningModule, TFLogger):
         padding = self.hparams.get("padding", 10)
         if self.hparams.get('visualize', False) or self.hparams.get("deploy", False):
             for i, (sample, pred) in enumerate(zip(batch, preds)):
+                # print(i, sample, pred)
                 instances = pred['instances']
                 boxes = instances.get('pred_boxes').tensor
                 class_id = instances.get('pred_classes')
@@ -94,6 +95,7 @@ class DetectionTask(pl.LightningModule, TFLogger):
                 boxes_nms = torch.clip(boxes_nms, 0, 640)
 
                 for j in range(len(scores_nms)):
+                    # print("- ", j)
                     instances = Instances((640, 640)) 
                     class_id_numpy = class_id_nms.to("cpu").numpy()[j]
                     box_numpy = boxes_nms.to("cpu").numpy()[j]
@@ -130,6 +132,7 @@ class DetectionTask(pl.LightningModule, TFLogger):
                         with open(json_save_path, 'w') as fp:
                                 json.dump(data, fp)
                     else:
+                        # print("save img", f"{self._output_save_dir}/{batch_nb}_{i}.jpg")
                         img_box.save(f"{self._output_save_dir}/{batch_nb}_{i}.jpg")
                     
         self.evaluator.process(batch, preds)
@@ -162,8 +165,12 @@ class DetectionTask(pl.LightningModule, TFLogger):
 
     def test_dataloader(self):
         if self.hparams.get('deploy', False):
-            dataset = load_dataset(self.hparams['dataset_name'])
-            df = pd.read_csv(self.hparams['deploy_meta_path']).query("downloaded == True")
+            dataset = get_dataset(None)#load_dataset(self.hparams['deploy_meta_path'])
+            # TODO: apprently it does ignore much of the loaded dataset
+            # can we just remove/restructure the 'load_dataset' call
+            df = pd.read_csv(self.hparams['deploy_meta_path'])
+            print(df)
+            df = df.query("downloaded == True")
             df["image_id"] = df['save_path']
             df["gsv_image_path"] = df['save_path']
             df['annotations'] = "[]"

detection/main.py

@@ -1,6 +1,7 @@
 import os
 import fire
 from pytorch_lightning import Trainer
+import torch
 
 from util.nni import run_nni
 from util import init_exp_folder, Args
@@ -110,6 +111,7 @@ def test(ckpt_path,
     trainer.test(task)
 
 
+
 def nni():
     run_nni(train, test)
 

requirements.txt

@@ -13,5 +13,5 @@ pytorch-ignite
 scikit-learn==0.23.2
 seaborn==0.10.1
 segmentation-models-pytorch
-torch==1.8.1+cu102 
-torchvision==0.9.1+cu102 
+torch
+torchvision