fix key not found; fix general_vfov_to_focal when batched input

Author: jinlinyi

demo/demo.py

@@ -162,4 +162,4 @@ def resize_fix_aspect_ratio(img, field, target_width=None, target_height=None):
 
 print("Alternatively, inference a batch of images")
 predictions = pf_model.inference_batch(img_bgr_list=[img_bgr, img_bgr, img_bgr])
-breakpoint()
+breakpoint()

perspective2d/modeling/param_network/param_network.py

@@ -210,15 +210,13 @@ def forward(self, predictions, batched_inputs=None):
                 param["pred_general_vfov"] = param["pred_vfov"]
             if "pred_rel_focal" not in param:
                 param["pred_rel_focal"] = torch.FloatTensor(
-                    [
-                        general_vfov_to_focal(
-                            to_numpy(param["pred_rel_cx"]),
-                            to_numpy(param["pred_rel_cy"]),
-                            1,
-                            to_numpy(param["pred_general_vfov"]),
-                            degree=True,
-                        )
-                    ]
+                    general_vfov_to_focal(
+                        to_numpy(param["pred_rel_cx"]),
+                        to_numpy(param["pred_rel_cy"]),
+                        1,
+                        to_numpy(param["pred_general_vfov"]),
+                        degree=True,
+                    )
                 )
             return param
 

perspective2d/perspectivefields.py

@@ -265,7 +265,7 @@ def forward(self, batched_inputs) -> dict:
                 param["pred_rel_cx"] = torch.zeros_like(param["pred_vfov"])
             if "pred_rel_cy" not in param.keys():
                 param["pred_rel_cy"] = torch.zeros_like(param["pred_vfov"])
-            assert len(processed_results) == len(param["pred_vfov"])
+            assert len(processed_results) == len(param["pred_general_vfov"])
             for i in range(len(processed_results)):
                 param_tmp = {k: v[i] for k, v in param.items()}
                 processed_results[i].update(param_tmp)

perspective2d/utils/utils.py

@@ -79,10 +79,14 @@ def fun(focal, *args):
         q_sqr = (focal / h) ** 2 + d_cx**2 + (d_cy - 0.5) ** 2
         cos_FoV = (p_sqr + q_sqr - 1) / 2 / np.sqrt(p_sqr) / np.sqrt(q_sqr)
         return cos_FoV - target_cos_FoV
-
     if degree:
         gvfov = np.radians(gvfov)
-    focal = scipy.optimize.fsolve(fun, 1.5, args=(h, rel_cx, rel_cy, np.cos(gvfov)))[0]
+    if type(rel_cx) != np.ndarray:
+        # if input is float
+        focal = scipy.optimize.fsolve(fun, 1.5, args=(h, rel_cx, rel_cy, np.cos(gvfov)))[0]
+    else:
+        # if input is numpy array
+        focal = scipy.optimize.fsolve(fun, np.ones(len(rel_cx)) * 1.5, args=(h, rel_cx, rel_cy, np.cos(gvfov)))
     focal = np.abs(focal)
     return focal