autoforge 1.7.0__tar.gz → 1.7.2__tar.gz

{autoforge-1.7.0 → autoforge-1.7.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: AutoForge
-Version: 1.7.0
+Version: 1.7.2
 Summary: AutoForge is a Python tool for generating 3D printed layered models from an input image.
 Author-email: Hendric Voss <hvoss@techfak.uni-bielefeld.de>
 License: CC BY-NC-SA 4.0
@@ -177,10 +177,29 @@ After running, the following files will be created in your specified output fold
 - **Hueforge Project File**: `project_file.hfp` 
 - **Swap Instructions**: `swap_instructions.txt`
 
+## Development
+
+To have a "nightly" version of the repository or have live updating changes during development please do the following: 
+
+```bash
+git clone https://github.com/hvoss-techfak/AutoForge.git
+cd AutoForge
+conda create -n forge python=3.11
+conda activate forge
+pip install -e .
+```
+
+If the installed pytorch version has no cuda support execute the following:
+
+```bash
+conda activate forge
+pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
+```
+
+
 ## Known Bugs
 
 - The optimizer can sometimes get stuck in a local minimum. If this happens, try running the optimization again with different settings.
-- Hueforge can't open STL files under linux. This is a known bug in hueforge.
 
 ## License
 

{autoforge-1.7.0 → autoforge-1.7.2}/README.md

@@ -151,10 +151,29 @@ After running, the following files will be created in your specified output fold
 - **Hueforge Project File**: `project_file.hfp` 
 - **Swap Instructions**: `swap_instructions.txt`
 
+## Development
+
+To have a "nightly" version of the repository or have live updating changes during development please do the following: 
+
+```bash
+git clone https://github.com/hvoss-techfak/AutoForge.git
+cd AutoForge
+conda create -n forge python=3.11
+conda activate forge
+pip install -e .
+```
+
+If the installed pytorch version has no cuda support execute the following:
+
+```bash
+conda activate forge
+pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
+```
+
+
 ## Known Bugs
 
 - The optimizer can sometimes get stuck in a local minimum. If this happens, try running the optimization again with different settings.
-- Hueforge can't open STL files under linux. This is a known bug in hueforge.
 
 ## License
 

{autoforge-1.7.0 → autoforge-1.7.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "AutoForge"
-version="1.7.0"
+version="1.7.2"
 description = "AutoForge is a Python tool for generating 3D printed layered models from an input image."
 authors = [{ name = "Hendric Voss", email = "hvoss@techfak.uni-bielefeld.de" }]
 readme = "README.md"

{autoforge-1.7.0 → autoforge-1.7.2}/src/AutoForge.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: AutoForge
-Version: 1.7.0
+Version: 1.7.2
 Summary: AutoForge is a Python tool for generating 3D printed layered models from an input image.
 Author-email: Hendric Voss <hvoss@techfak.uni-bielefeld.de>
 License: CC BY-NC-SA 4.0
@@ -177,10 +177,29 @@ After running, the following files will be created in your specified output fold
 - **Hueforge Project File**: `project_file.hfp` 
 - **Swap Instructions**: `swap_instructions.txt`
 
+## Development
+
+To have a "nightly" version of the repository or have live updating changes during development please do the following: 
+
+```bash
+git clone https://github.com/hvoss-techfak/AutoForge.git
+cd AutoForge
+conda create -n forge python=3.11
+conda activate forge
+pip install -e .
+```
+
+If the installed pytorch version has no cuda support execute the following:
+
+```bash
+conda activate forge
+pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
+```
+
+
 ## Known Bugs
 
 - The optimizer can sometimes get stuck in a local minimum. If this happens, try running the optimization again with different settings.
-- Hueforge can't open STL files under linux. This is a known bug in hueforge.
 
 ## License
 

{autoforge-1.7.0 → autoforge-1.7.2}/src/autoforge/Helper/ImageHelper.py

@@ -1,5 +1,21 @@
+import os
+from typing import Sequence
+
 import cv2
 import torch
+from cv2.typing import MatLike
+import numpy as np
+
+
+def imread(filename: str, flags: int = cv2.IMREAD_COLOR) -> MatLike:
+    return cv2.imdecode(np.fromfile(filename, dtype=np.uint8), flags)
+
+
+def imwrite(filename: str, img: MatLike, params: Sequence[int] = ()) -> None:
+    success, encoded_img = cv2.imencode(os.path.splitext(filename)[1], img, params)
+    if not success:
+        raise OSError(f"cv2 could not write to path {filename}")
+    encoded_img.tofile(filename)
 
 
 def resize_image(img, max_size):

{autoforge-1.7.0 → autoforge-1.7.2}/src/autoforge/Helper/OutputHelper.py

@@ -98,12 +98,14 @@ def generate_project_file(
             "Color": mat[" Color"],
             "Name": mat[" Name"],
             # Convert Owned to a boolean (in case it is read as a string)
-            "Owned": str(mat[" Owned"]).strip().lower() == "true",
+            "Owned": str(mat[" Owned"]).strip().lower() == "true"
+            if " Owned" in mat
+            else False,
             "Transmissivity": float(mat[" TD"])
             if not float(mat[" TD"]).is_integer()
             else int(mat[" TD"]),
-            "Type": mat[" Type"],
-            "uuid": mat[" Uuid"],
+            "Type": mat[" Type"] if " Type" in mat else "PLA",
+            "uuid": mat[" Uuid"] if " Uuid" in mat else str(uuid.uuid4()),
         }
         filament_set.append(filament_entry)
 

{autoforge-1.7.0 → autoforge-1.7.2}/src/autoforge/Modules/Optimizer.py

@@ -5,12 +5,6 @@ import matplotlib.pyplot as plt
 import numpy as np
 import torch
 from torch.utils.tensorboard import SummaryWriter
-from torchcam.methods import SmoothGradCAMpp
-from torchcam.utils import overlay_mask
-from torchvision import models
-from torchvision.models import ResNet18_Weights
-from torchvision.transforms._functional_tensor import resize, normalize
-from torchvision.transforms.functional import to_pil_image
 from tqdm import tqdm
 
 from autoforge.Helper.CAdamW import CAdamW
@@ -119,48 +113,48 @@ class FilamentOptimizer:
             lr=self.learning_rate,
         )
 
-        # Setup the focus map component
-        # Load a state-of-the-art ResNet50 model with pretrained weights.
-        cam_model = models.resnet18(weights=ResNet18_Weights.DEFAULT).to(
-            self.target.device
-        )
-        cam_model.eval()
-        # Use ScoreCAM from torchcam on the 'layer4' block.
-        cam_extractor = SmoothGradCAMpp(cam_model, target_layer="layer4")
-        # Preprocess the target image for ResNet50.
-        target_preprocessed = normalize(
-            resize(self.target.permute(2, 0, 1), (224, 224)) / 255.0,
-            [0.485, 0.456, 0.406],
-            [0.229, 0.224, 0.225],
-        )
-        focus_maps = []
-        for _ in tqdm(range(1), desc="Creating loss function attention map..."):
-            out = cam_model(target_preprocessed.unsqueeze(0))
-            # ScoreCAM returns a dict with keys corresponding to layers.
-            focus_map = cam_extractor(out.squeeze(0).argmax().item(), out)[0]
-            # Normalize the focus map to [0, 1]
-            focus_map = (focus_map - focus_map.min()) / (
-                focus_map.max() - focus_map.min() + 1e-8
-            )
-            focus_maps.append(focus_map)
-        # get maximum over all maps
-        self.focus_map = torch.stack(focus_maps)
-        self.focus_map = self.focus_map.max(dim=0)[0]
-        # normalize again
-        self.focus_map = (self.focus_map - self.focus_map.min()) / (
-            self.focus_map.max() - self.focus_map.min() + 1e-8
-        )
-
-        # resize back to target size, current dim is [1,h,w]
-        self.focus_map = (
-            torch.nn.functional.interpolate(
-                self.focus_map.unsqueeze(0),
-                (self.H, self.W),
-                mode="bilinear",
-            )
-            .squeeze(0)
-            .squeeze(0)
-        )
+        # # Setup the focus map component
+        # # Load a state-of-the-art ResNet50 model with pretrained weights.
+        # cam_model = models.resnet18(weights=ResNet18_Weights.DEFAULT).to(
+        #     self.target.device
+        # )
+        # cam_model.eval()
+        # # Use ScoreCAM from torchcam on the 'layer4' block.
+        # cam_extractor = SmoothGradCAMpp(cam_model, target_layer="layer4")
+        # # Preprocess the target image for ResNet50.
+        # target_preprocessed = normalize(
+        #     resize(self.target.permute(2, 0, 1), (224, 224)) / 255.0,
+        #     [0.485, 0.456, 0.406],
+        #     [0.229, 0.224, 0.225],
+        # )
+        # focus_maps = []
+        # for _ in tqdm(range(1), desc="Creating loss function attention map..."):
+        #     out = cam_model(target_preprocessed.unsqueeze(0))
+        #     # ScoreCAM returns a dict with keys corresponding to layers.
+        #     focus_map = cam_extractor(out.squeeze(0).argmax().item(), out)[0]
+        #     # Normalize the focus map to [0, 1]
+        #     focus_map = (focus_map - focus_map.min()) / (
+        #         focus_map.max() - focus_map.min() + 1e-8
+        #     )
+        #     focus_maps.append(focus_map)
+        # # get maximum over all maps
+        # self.focus_map = torch.stack(focus_maps)
+        # self.focus_map = self.focus_map.max(dim=0)[0]
+        # # normalize again
+        # self.focus_map = (self.focus_map - self.focus_map.min()) / (
+        #     self.focus_map.max() - self.focus_map.min() + 1e-8
+        # )
+        #
+        # # resize back to target size, current dim is [1,h,w]
+        # self.focus_map = (
+        #     torch.nn.functional.interpolate(
+        #         self.focus_map.unsqueeze(0),
+        #         (self.H, self.W),
+        #         mode="bilinear",
+        #     )
+        #     .squeeze(0)
+        #     .squeeze(0)
+        # )
 
         # Setup best discrete solution tracking
         self.best_discrete_loss = float("inf")
@@ -171,7 +165,8 @@ class FilamentOptimizer:
 
         # If you want a figure for real-time visualization:
         if self.visualize_flag:
-            plt.ion()
+            if self.args.disable_visualization_for_gradio != 1:
+                plt.ion()
             self.fig, self.ax = plt.subplots(2, 3, figsize=(14, 6))
 
             self.target_im_ax = self.ax[0, 0].imshow(
@@ -188,7 +183,8 @@ class FilamentOptimizer:
                 np.zeros((self.H, self.W, 3), dtype=np.uint8)
             )
             self.ax[0, 2].set_title("Best Discrete Composite")
-            plt.pause(0.1)
+            if self.args.disable_visualization_for_gradio != 1:
+                plt.pause(0.1)
 
             self.depth_map_ax = self.ax[1, 0].imshow(
                 np.zeros((self.H, self.W), dtype=np.uint8), cmap="viridis"
@@ -200,17 +196,14 @@ class FilamentOptimizer:
             )
             self.ax[1, 1].set_title("Height Map Changes")
 
-            target_vis = self.target.cpu().detach().numpy()
-            focus_vis = self.focus_map.cpu().detach().numpy()
-
-            overlay = overlay_mask(
-                to_pil_image(target_vis),
-                to_pil_image(focus_vis, mode="F"),
-                alpha=0.5,
-            )
-
-            self.focus_map_ax = self.ax[1, 2].imshow(overlay)
-            self.ax[1, 2].set_title("Image Focus Map (currently unused)")
+            # target_vis = self.target.cpu().detach().numpy()
+            # focus_vis = self.focus_map.cpu().detach().numpy()
+            #
+            # overlay = overlay_mask(
+            #     to_pil_image(target_vis),
+            #     to_pil_image(focus_vis, mode="F"),
+            #     alpha=0.5,
+            # )
 
             # Compute and store the initial height map for later difference computation.
             with torch.no_grad():
@@ -284,7 +277,7 @@ class FilamentOptimizer:
             material_TDs=self.material_TDs,
             background=self.background,
             add_penalty_loss=10.0,
-            focus_map=self.focus_map,
+            focus_map=None,
             focus_strength=0.0,
         )
 
@@ -445,10 +438,11 @@ class FilamentOptimizer:
         self.diff_depth_map_ax.set_clim(-2.5, 2.5)
 
         self.fig.suptitle(
-            f"Step {self.num_steps_done}, Tau: {tau_g:.4f}, Loss: {self.loss:.4f}, Best Discrete Loss: {self.best_discrete_loss:.4f}"
+            f"Step {self.num_steps_done}/{self.args.iterations}, Tau: {tau_g:.4f}, Loss: {self.loss:.4f}, Best Discrete Loss: {self.best_discrete_loss:.4f}"
         )
-
-        plt.pause(0.01)
+        if self.args.disable_visualization_for_gradio != 1:
+            plt.pause(0.01)
+        plt.savefig(self.args.output_folder + "/vis_temp.png")
 
     def get_current_parameters(self):
         """
@@ -544,8 +538,9 @@ class FilamentOptimizer:
             prune_num_swaps,
             prune_redundant_layers,
         )
+
         if search_seed:
-            self.rng_seed_search(self.best_discrete_loss, 1000, autoset_seed=True)
+            self.rng_seed_search(self.best_discrete_loss, 100, autoset_seed=True)
 
         prune_num_colors(
             self,
@@ -605,9 +600,6 @@ class FilamentOptimizer:
                     rng_seed=seed,
                 )
 
-            # 3) Compute the "discrete" loss function
-            #    If you have a separate function for that, just call it.
-            #    Otherwise, reuse the standard compute_loss with add_penalty_loss=False, etc.
             current_disc_loss = compute_loss(
                 comp=comp_disc,
                 target=self.target,

{autoforge-1.7.0 → autoforge-1.7.2}/src/autoforge/auto_forge.py

@@ -1,3 +1,4 @@
+import argparse
 import sys
 import os
 import time
@@ -15,7 +16,7 @@ from autoforge.Helper.Heightmaps.DepthEstimateHeightMap import (
     init_height_map_depth_color_adjusted,
 )
 
-from autoforge.Helper.ImageHelper import resize_image
+from autoforge.Helper.ImageHelper import resize_image, imread
 from autoforge.Helper.OutputHelper import (
     generate_stl,
     generate_swap_instructions,
@@ -120,6 +121,7 @@ def main():
         type=bool,
         default=True,
         help="Enable visualization during optimization",
+        action=argparse.BooleanOptionalAction,
     )
 
     # Instead of an output_size parameter, we use stl_output_size and nozzle_diameter.
@@ -147,12 +149,14 @@ def main():
         type=bool,
         default=True,
         help="Perform pruning after optimization",
+        action=argparse.BooleanOptionalAction,
     )
     parser.add_argument(
         "--fast_pruning",
         type=bool,
         default=True,
         help="Use fast pruning method",
+        action=argparse.BooleanOptionalAction,
     )
     parser.add_argument(
         "--fast_pruning_percent",
@@ -242,7 +246,7 @@ def main():
     parser.add_argument(
         "--num_init_rounds",
         type=int,
-        default=64,
+        default=128,
         help="Number of rounds to choose the starting height map from.",
     )
     parser.add_argument(
@@ -251,6 +255,12 @@ def main():
         default=-1,
         help="Number of layers to cluster the image into.",
     )
+    parser.add_argument(
+        "--disable_visualization_for_gradio",
+        type=int,
+        default=0,
+        help="Simple switch to disable the matplotlib render window for gradio rendering.",
+    )
 
     args = parser.parse_args()
     if args.num_init_cluster_layers == -1:
@@ -302,7 +312,7 @@ def main():
     material_TDs = torch.tensor(material_TDs_np, dtype=torch.float32, device=device)
 
     # Read input image
-    img = cv2.imread(args.input_image, cv2.IMREAD_UNCHANGED)
+    img = imread(args.input_image, cv2.IMREAD_UNCHANGED)
     computed_output_size = int(round(args.stl_output_size * 2 / args.nozzle_diameter))
     print(f"Computed solving pixel size: {computed_output_size}")
     alpha = None
@@ -380,7 +390,7 @@ def main():
         for i in tbar:
             loss_val = optimizer.step(record_best=i % 8 == 0)
 
-            optimizer.visualize(interval=25)
+            optimizer.visualize(interval=100)
             optimizer.log_to_tensorboard(interval=100)
 
             if (i + 1) % 100 == 0: