add camera position visualization

Author: mks0601

MANO_render/J_regressor_mano_ih26m.npy …l/MANO_render/J_regressor_mano_ih26m.npyMANO_render/J_regressor_mano_ih26m.npy renamed to tool/MANO_render/J_regressor_mano_ih26m.npy MANO_render/J_regressor_mano_ih26m.npy renamed to tool/MANO_render/J_regressor_mano_ih26m.npy

@@ -61,7 +61,7 @@ def get_fitting_error(mesh, regressor, cam_params, joints, hand_type, capture_id
     print('Fix shapedirs bug of MANO')
     mano_layer['left'].shapedirs[:,0,:] *= -1
 
-root_path = 'PATH_TO_DB'
+root_path = '../../data/InterHand2.6M/data/'
 img_root_path = osp.join(root_path, 'images')
 annot_root_path = osp.join(root_path, 'annotations')
 subset = 'all'

MANO_render/render.py tool/MANO_render/render.pyMANO_render/render.py renamed to tool/MANO_render/render.py

@@ -0,0 +1,44 @@
+import json
+import numpy as np
+from mpl_toolkits.mplot3d import Axes3D
+import matplotlib.pyplot as plt
+import matplotlib as mpl
+
+# load camera positions
+campos_dict = {}
+for subset in ('all', 'machine_annot', 'human_annot'):
+    for split in ('train', 'test', 'val'):
+        
+        if split == 'val' and subset != 'machine_annot':
+            continue
+        
+        with open('../../data/InterHand2.6M/data/annotations/' + subset + '/InterHand2.6M_' + split + '_camera.json') as f:
+            cameras = json.load(f)
+        for capture_id in cameras.keys():
+            for cam_id in cameras[capture_id]['campos'].keys():
+                campos = np.array(cameras[capture_id]['campos'][cam_id], dtype=np.float32)
+
+                # exact camera positions can be slightly different for each 'capture_id'.
+                # however, just overwrite them for the visualization purpose.
+                if cam_id not in campos_dict:
+                    campos_dict[cam_id] = campos
+
+# plot camera positions
+fig = plt.figure()
+ax = fig.add_subplot(111, projection='3d')
+
+# Convert from plt 0-1 RGBA colors to 0-255 BGR colors for opencv.
+cmap = plt.get_cmap('rainbow')
+cam_num = len(campos_dict)
+colors = [cmap(i) for i in np.linspace(0, 1, cam_num)]
+colors = [np.array((c[2], c[1], c[0])) for c in colors]
+
+for i, k in enumerate(campos_dict.keys()):
+    ax.scatter(campos_dict[k][0], campos_dict[k][2], -campos_dict[k][1], c=colors[i], marker='o')
+    ax.text(campos_dict[k][0], campos_dict[k][2], -campos_dict[k][1], k)
+
+ax.set_xticklabels([])
+ax.set_yticklabels([])
+ax.set_zticklabels([])
+plt.show()
+