stuck on singleview visualization

Hi，I have run the run_cosypose_eval.py and get the 'results/xxxx/results.pth.tar', then I want to visualize the output so I run the visualize_singleview_predictions.ipynb but the html picture I get there is no objects rendered. I print the TCO.

Setting OMP and MKL num threads to 1. pybullet build time: Nov 18 2020 10:41:54 datasets_cfg.py tless.primesense.test 0:00:00.699286 - Building index and loading annotations... [Memory]1.0s, 0.0min : Loading build_index... self.all_labels ['obj_101', 'obj_102', 'obj_103', 'obj_104', 'obj_105', 'obj_107', 'obj_108', 'obj_109', 'obj_110', 'obj_111', 'obj_112'] EGL device choice: 0 of 10 (from EGL_VISIBLE_DEVICES) Loaded EGL 1.5 after reload. GL_VENDOR=NVIDIA Corporation GL_RENDERER=TITAN RTX/PCIe/SSE2 GL_VERSION=4.6.0 NVIDIA 450.80.02 GL_SHADING_LANGUAGE_VERSION=4.60 NVIDIA Version = 4.6.0 NVIDIA 450.80.02 Vendor = NVIDIA Corporation Renderer = TITAN RTX/PCIe/SSE2 TWO R = 0.999047 -0.0096753 -0.0425613 -0.028929 -0.876956 -0.4797 -0.0326831 0.480474 -0.8764 p = -0.624104 -1.52404 2.19855

TWO R = 0.999572 -0.00120914 -0.0292267 -0.0279897 -0.329852 -0.943618 -0.00849952 0.944032 -0.329744 p = -0.44151 -1.62043 2.61096

TWO R = 0.999217 -0.0169255 -0.0357491 -0.0346052 -0.811816 -0.582887 -0.019156 0.583668 -0.811767 p = -0.57851 -0.982422 2.13662

TWO R = 0.999214 -0.0152667 -0.0365939 -0.032582 -0.842089 -0.538354 -0.0225965 0.539123 -0.841924 p = -0.726713 -0.688979 2.10162

TWO R = 0.999357 -0.00802674 -0.034947 -0.0297393 -0.73004 -0.682757 -0.0200324 0.683358 -0.729809 p = -0.818724 -1.20954 2.21362

TWO R = 0.999251 -0.0148427 -0.0357447 -0.0338669 -0.78238 -0.62188 -0.0187355 0.622624 -0.782296 p = -0.970052 -0.791749 2.15384

can someone help me? by the way, I found the code in bullet_scene_renderer.py, are those objects rendered through this code?:

def render_scene(self, obj_infos, cam_infos, render_depth=False):
    self.setup_scene(obj_infos)
    return self.render_images(cam_infos, render_depth=render_depth)

def setup_scene(self, obj_infos):
    labels = [obj['name'] for obj in obj_infos]
    bodies = self.body_cache.get_bodies_by_labels(labels)

    for (obj_info, body) in zip(obj_infos, bodies):
        TWO = Transform(obj_info['TWO'])
        body.pose = TWO
        print('TWO', TWO)
        color = obj_info.get('color', None)
        if color is not None:
            pb.changeVisualShape(body.body_id, -1, physicsClientId=0, rgbaColor=color)
    return bodies

but the returned bodies never used, I wonder how does it work? by passing address? many thanks!

ylabbe / cosypose

stuck on singleview visualization #36