add control glpn depth processor

modules/control/proc/dpt.py

@@ -31,4 +31,5 @@ class DPTDetector:
             output = prediction.squeeze().cpu().numpy()
             formatted = (output * 255 / np.max(output)).astype("uint8")
             depth = Image.fromarray(formatted)
+            depth = depth.convert('RGB')
             return depth

modules/control/proc/glpn.py

@@ -0,0 +1,35 @@
+from PIL import Image
+import numpy as np
+import torch
+from transformers import AutoImageProcessor, GLPNForDepthEstimation
+from modules import devices
+
+
+image_processor: AutoImageProcessor = None
+glpn_model: GLPNForDepthEstimation = None
+
+
+class GLPNDetector:
+    def __call__(self, input_image=None):
+        global image_processor, glpn_model # pylint: disable=global-statement
+        from modules.control.processors import cache_dir
+        if image_processor is None:
+            image_processor = AutoImageProcessor.from_pretrained("vinvino02/glpn-kitti", cache_dir=cache_dir)
+        if glpn_model is None:
+            glpn_model = GLPNForDepthEstimation.from_pretrained("vinvino02/glpn-kitti", cache_dir=cache_dir)
+
+        with devices.inference_context():
+            inputs = image_processor(images=input_image, return_tensors="pt")
+            outputs = glpn_model(**inputs)
+            predicted_depth = outputs.predicted_depth
+            prediction = torch.nn.functional.interpolate(
+                predicted_depth.unsqueeze(1),
+                size=input_image.size[::-1],
+                mode="bicubic",
+                align_corners=False,
+            )
+            output = prediction.squeeze().cpu().numpy()
+            formatted = 255 - (output * 255 / np.max(output)).astype("uint8")
+            depth = Image.fromarray(formatted)
+            depth = depth.convert('RGB')
+            return depth

modules/control/processors.py

@@ -24,6 +24,7 @@ from modules.control.proc.segment_anything import SamDetector
 from modules.control.proc.zoe import ZoeDetector
 from modules.control.proc.marigold import MarigoldDetector
 from modules.control.proc.dpt import DPTDetector
+from modules.control.proc.glpn import GLPNDetector
 
 
 models = {}
@@ -54,6 +55,7 @@ config = {
     'MLSD': {'class': MLSDdetector, 'checkpoint': True, 'params': {'thr_v': 0.1, 'thr_d': 0.1}},
     'Shuffle': {'class': ContentShuffleDetector, 'checkpoint': False, 'params': {}},
     'DPT Depth Hybrid': {'class': DPTDetector, 'checkpoint': False, 'params': {}},
+    'GLPN Depth': {'class': GLPNDetector, 'checkpoint': False, 'params': {}},
     # 'Midas Depth Large': {'class': MidasDetector, 'checkpoint': True, 'params': {'bg_th': 0.1, 'depth_and_normal': False}, 'load_config': {'pretrained_model_or_path': 'Intel/dpt-large', 'model_type': "dpt_large", 'filename': ''}},
     # 'Zoe Depth Zoe': {'class': ZoeDetector, 'checkpoint': True, 'params': {}},
     # 'Zoe Depth NK': {'class': ZoeDetector, 'checkpoint': True, 'params': {}, 'load_config': {'pretrained_model_or_path': 'halffried/gyre_zoedepth', 'filename': 'ZoeD_M12_NK.safetensors', 'model_type': "zoedepth_nk"}},