Merge pull request #291 from leggedrobotics/dev/open_source_jf

enjoy reviewing the pull request

.deprecated/dataset/__init__.py

@@ -0,0 +1 @@
+from .graph_trav_dataset import get_ablation_module

wild_visual_navigation/dataset/graph_trav_dataset.py → .deprecated/dataset/graph_trav_dataset.py

@@ -1,14 +1,13 @@
-from torch_geometric.data import InMemoryDataset, DataListLoader, DataLoader
-from torch_geometric.data import LightningDataset
+# TODO
+# from torch_geometric.data import InMemoryDataset, DataLoader
+# from torch_geometric.data import LightningDataset
+# from torch_geometric.data import Dataset
 
 from wild_visual_navigation import WVN_ROOT_DIR
 import os
 import torch
-from pathlib import Path
-from torch_geometric.data import Dataset
 from torchvision import transforms as T
 from typing import Optional, Callable
-from torch_geometric.data import Data
 import random
 
 
@@ -186,13 +185,28 @@ def get_test_dataset(perugia_root, env, feature_key, test_all_datasets, training
                 )
             ]
         else:
-            val_dataset = get_test_dataset(perugia_root, env, feature_key, test_all_datasets, training_data_percentage)
+            val_dataset = get_test_dataset(
+                perugia_root,
+                env,
+                feature_key,
+                test_all_datasets,
+                training_data_percentage,
+            )
 
         train_loader = DataLoader(
-            dataset=train_dataset, batch_size=batch_size, num_workers=num_workers, pin_memory=False
+            dataset=train_dataset,
+            batch_size=batch_size,
+            num_workers=num_workers,
+            pin_memory=False,
         )
         val_loader = [
-            DataLoader(dataset=v, batch_size=batch_size, num_workers=num_workers, pin_memory=False) for v in val_dataset
+            DataLoader(
+                dataset=v,
+                batch_size=batch_size,
+                num_workers=num_workers,
+                pin_memory=False,
+            )
+            for v in val_dataset
         ]
 
     else:
@@ -212,9 +226,20 @@ def get_test_dataset(perugia_root, env, feature_key, test_all_datasets, training
                 )
             ]
         else:
-            test_dataset = get_test_dataset(perugia_root, env, feature_key, test_all_datasets, training_data_percentage)
+            test_dataset = get_test_dataset(
+                perugia_root,
+                env,
+                feature_key,
+                test_all_datasets,
+                training_data_percentage,
+            )
         test_loader = [
-            DataLoader(dataset=t, batch_size=batch_size, num_workers=num_workers, pin_memory=False)
+            DataLoader(
+                dataset=t,
+                batch_size=batch_size,
+                num_workers=num_workers,
+                pin_memory=False,
+            )
             for t in test_dataset
         ]
     else:

wild_visual_navigation/general/training_routine.py → .deprecated/general/training_routine.py

@@ -7,7 +7,6 @@
 
 # Frameworks
 import torch
-import pytorch_lightning
 from pytorch_lightning import seed_everything, Trainer
 from pytorch_lightning.callbacks import EarlyStopping
 from pytorch_lightning.callbacks import ModelCheckpoint
@@ -60,7 +59,11 @@ def training_routine(exp: ExperimentParams, seed=42) -> torch.Tensor:
 
     if exp.cb_checkpoint.active:
         checkpoint_callback = ModelCheckpoint(
-            dirpath=model_path, save_top_k=1, monitor="epoch", mode="max", save_last=True
+            dirpath=model_path,
+            save_top_k=1,
+            monitor="epoch",
+            mode="max",
+            save_last=True,
         )
         cb_ls.append(checkpoint_callback)
 
@@ -93,7 +96,7 @@ def training_routine(exp: ExperimentParams, seed=42) -> torch.Tensor:
         ckpt = torch.load(exp.model.load_ckpt)
         try:
             res = model.load_state_dict(ckpt.state_dict, strict=False)
-        except:
+        except Exception:
             res = model.load_state_dict(ckpt, strict=False)
         print("Loaded model checkpoint:", res)
     trainer = Trainer(**exp.trainer, callbacks=cb_ls, logger=logger)

wild_visual_navigation/lightning/lightning_module.py → .deprecated/lightning/lightning_module.py

@@ -2,15 +2,14 @@
 import pytorch_lightning as pl
 import torch
 import torch.nn.functional as F
-from os.path import join
 from wild_visual_navigation.visu import LearningVisualizer
 from pytorch_lightning.utilities.types import EPOCH_OUTPUT
-from torch_geometric.data import Data
+
+# from torch_geometric.data import Data
 from torchmetrics import ROC
 
 from wild_visual_navigation.utils import TraversabilityLoss, MetricLogger
 import os
-import pickle
 
 
 class LightningTrav(pl.LightningModule):
@@ -70,8 +69,20 @@ def training_step(self, batch: any, batch_idx: int) -> torch.Tensor:
 
         for k, v in loss_aux.items():
             if k.find("loss") != -1:
-                self.log(f"{self._mode}_{k}", v.item(), on_epoch=True, prog_bar=True, batch_size=BS)
-        self.log(f"{self._mode}_loss", loss.item(), on_epoch=True, prog_bar=True, batch_size=BS)
+                self.log(
+                    f"{self._mode}_{k}",
+                    v.item(),
+                    on_epoch=True,
+                    prog_bar=True,
+                    batch_size=BS,
+                )
+        self.log(
+            f"{self._mode}_loss",
+            loss.item(),
+            on_epoch=True,
+            prog_bar=True,
+            batch_size=BS,
+        )
 
         self.visu(graph, res_updated, loss_aux["confidence"])
 
@@ -112,9 +123,21 @@ def validation_step(self, batch: any, batch_idx: int, dataloader_id: int = 0) ->
 
         for k, v in loss_aux.items():
             if k.find("loss") != -1:
-                self.log(f"{self._mode}_{k}", v.item(), on_epoch=True, prog_bar=True, batch_size=BS)
+                self.log(
+                    f"{self._mode}_{k}",
+                    v.item(),
+                    on_epoch=True,
+                    prog_bar=True,
+                    batch_size=BS,
+                )
 
-        self.log(f"{self._mode}_loss", loss.item(), on_epoch=True, prog_bar=True, batch_size=BS)
+        self.log(
+            f"{self._mode}_loss",
+            loss.item(),
+            on_epoch=True,
+            prog_bar=True,
+            batch_size=BS,
+        )
         self.visu(graph, res_updated, loss_aux["confidence"])
 
         return loss
@@ -133,7 +156,7 @@ def update_threshold(self):
                 fpr, tpr, thresholds = self._auxiliary_training_roc.compute()
                 index = torch.where(fpr > 0.15)[0][0]
                 self.threshold[0] = thresholds[index]
-            except:
+            except Exception:
                 pass
         else:
             self.threshold[0] = 0.5
@@ -158,13 +181,25 @@ def test_step(self, batch: any, batch_idx: int, dataloader_id: int = 0) -> torch
 
         for k, v in loss_aux.items():
             if k.find("loss") != -1:
-                self.log(f"{self._mode}_{k}", v.item(), on_epoch=True, prog_bar=True, batch_size=BS)
-        self.log(f"{self._mode}_loss", loss.item(), on_epoch=True, prog_bar=True, batch_size=BS)
+                self.log(
+                    f"{self._mode}_{k}",
+                    v.item(),
+                    on_epoch=True,
+                    prog_bar=True,
+                    batch_size=BS,
+                )
+        self.log(
+            f"{self._mode}_loss",
+            loss.item(),
+            on_epoch=True,
+            prog_bar=True,
+            batch_size=BS,
+        )
         self.visu(graph, res_updated, loss_aux["confidence"])
         return loss
 
     def test_epoch_end(self, outputs: any, plot=False):
-        ################ NEW VERSION ################
+        # NEW VERSION
         res = self._metric_logger.get_epoch_results("test")
         dic2 = {
             "trainer_logged_metric" + k: v.item()
@@ -199,7 +234,7 @@ def visu(self, graph: Data, res: torch.tensor, confidence: torch.tensor):
                 on_epoch=True,
                 batch_size=graph.ptr.shape[0] - 1,
             )
-        except:
+        except Exception:
             pass
 
         for b in range(graph.ptr.shape[0] - 1):
@@ -232,14 +267,22 @@ def visu(self, graph: Data, res: torch.tensor, confidence: torch.tensor):
 
                 # Visualize Graph with Segmentation
                 t1 = self._visualizer.plot_traversability_graph_on_seg(
-                    pred[:, 0], seg, graph[b], center, img, not_log=True, colorize_invalid_centers=True
+                    pred[:, 0],
+                    seg,
+                    graph[b],
+                    center,
+                    img,
+                    not_log=True,
+                    colorize_invalid_centers=True,
                 )
                 t2 = self._visualizer.plot_traversability_graph_on_seg(
                     graph[b].y, seg, graph[b], center, img, not_log=True
                 )
                 t_img = self._visualizer.plot_image(img, not_log=True)
                 self._visualizer.plot_list(
-                    imgs=[t1, t2, t_img], tag=f"C{c}_{self._mode}_GraphTrav", store_folder=f"{self._mode}/graph_trav"
+                    imgs=[t1, t2, t_img],
+                    tag=f"C{c}_{self._mode}_GraphTrav",
+                    store_folder=f"{self._mode}/graph_trav",
                 )
 
                 nr_channel_reco = graph[b].x.shape[1]
@@ -248,14 +291,27 @@ def visu(self, graph: Data, res: torch.tensor, confidence: torch.tensor):
 
                 # # Visualize Graph with Segmentation
                 c1 = self._visualizer.plot_traversability_graph_on_seg(
-                    conf, seg, graph[b], center, img, not_log=True, colorize_invalid_centers=True
+                    conf,
+                    seg,
+                    graph[b],
+                    center,
+                    img,
+                    not_log=True,
+                    colorize_invalid_centers=True,
                 )
                 c2 = self._visualizer.plot_traversability_graph_on_seg(
-                    graph[b].y_valid.type(torch.float32), seg, graph[b], center, img, not_log=True
+                    graph[b].y_valid.type(torch.float32),
+                    seg,
+                    graph[b],
+                    center,
+                    img,
+                    not_log=True,
                 )
                 c_img = self._visualizer.plot_image(img, not_log=True)
                 self._visualizer.plot_list(
-                    imgs=[c1, c2, c_img], tag=f"C{c}_{self._mode}_GraphConf", store_folder=f"{self._mode}/graph_conf"
+                    imgs=[c1, c2, c_img],
+                    tag=f"C{c}_{self._mode}_GraphConf",
+                    store_folder=f"{self._mode}/graph_conf",
                 )
 
                 # if self._mode == "test":
@@ -316,10 +372,18 @@ def visu(self, graph: Data, res: torch.tensor, confidence: torch.tensor):
             reco_loss = F.mse_loss(pred[:, -nr_channel_reco:], graph[b].x, reduction="none").mean(dim=1)
 
             self._visualizer.plot_histogram(
-                reco_loss, graph[b].y, mean, std, tag=f"C{c}_{self._mode}__confidence_generator_prop"
+                reco_loss,
+                graph[b].y,
+                mean,
+                std,
+                tag=f"C{c}_{self._mode}__confidence_generator_prop",
             )
 
             if hasattr(graph[b], "y_gt"):
                 self._visualizer.plot_histogram(
-                    reco_loss, graph[b].y_gt, mean, std, tag=f"C{c}_{self._mode}__confidence_generator_gt"
+                    reco_loss,
+                    graph[b].y_gt,
+                    mean,
+                    std,
+                    tag=f"C{c}_{self._mode}__confidence_generator_gt",
                 )

scripts/ablations/sklearn_ablation.py → .deprecated/scripts/ablations/sklearn_ablation.py

@@ -1,7 +1,6 @@
 raise ValueError("TODO: Not tested with new configuration!")
 # Use the same config to load the data using the dataloader
 from wild_visual_navigation.dataset import get_ablation_module
-from wild_visual_navigation import WVN_ROOT_DIR
 from wild_visual_navigation.cfg import ExperimentParams
 import torch
 from torchmetrics import Accuracy, AUROC
@@ -10,10 +9,8 @@
 
 patch_sklearn()
 
-from sklearn.neighbors import KNeighborsClassifier
 from sklearn.svm import SVC
 from sklearn.ensemble import RandomForestClassifier
-from sklearn.neural_network import MLPClassifier
 import numpy as np
 import copy
 import os
@@ -125,8 +122,14 @@
 
             for features_test, y_gt, y_prop, test_scene in zip(features_tests, y_gts, y_props, test_scenes):
                 y_pred = v.predict_proba(features_test)
-                test_auroc_gt.update(torch.from_numpy(y_pred[:, 1]), torch.from_numpy(y_gt).type(torch.long))
-                test_auroc_prop.update(torch.from_numpy(y_pred[:, 1]), torch.from_numpy(y_prop).type(torch.long))
+                test_auroc_gt.update(
+                    torch.from_numpy(y_pred[:, 1]),
+                    torch.from_numpy(y_gt).type(torch.long),
+                )
+                test_auroc_prop.update(
+                    torch.from_numpy(y_pred[:, 1]),
+                    torch.from_numpy(y_prop).type(torch.long),
+                )
 
                 res = {
                     "test_auroc_gt_seg": test_auroc_gt.compute().item(),
@@ -146,15 +149,16 @@
     ws = os.environ.get("ENV_WORKSTATION_NAME", "default")
     # Store epoch output to disk.
     p = os.path.join(
-        exp.env.results, f"ablations/classicial_learning_ablation_{ws}/classicial_learning_ablation_test_results.pkl"
+        exp.env.results,
+        f"ablations/classicial_learning_ablation_{ws}/classicial_learning_ablation_test_results.pkl",
     )
     print(results_epoch)
     Path(p).parent.mkdir(parents=True, exist_ok=True)
 
     try:
         os.remove(p)
     except OSError as error:
-        pass
+        print(error)
 
     with open(p, "wb") as handle:
         pickle.dump(results_epoch, handle, protocol=pickle.HIGHEST_PROTOCOL)