[feature request] More Guidance of Evaluating with ICP

[DateBro]

Hi, I want to evaluate the T-LESS dataset using the AAE with ICP and I have read #26, which still makes me confused about how to evaluate with ICP. Could you give some more finer details about the usage of ICP, like what should I set in aae_retina_webcam.cfg and eval_group/eval_my_autoencoder.cfg? I tried evaluating the AAE with the following aae_retina_webcam.cfg and got a Segmentation Fault.

[MODEL]
gpu_memory_fraction = 0.7

[DATA]
color_format = bgr
color_data_type = np.float32
depth_data_type = np.float32
depth_scale = 1000.

[AAE]
experiments = ['tless/nobn']
upright = False
topk = 1

[DETECTOR]
detector_model_path = /home/zhiyong/server/GithubFiles/keras-retinanet/snapshots/img04_batch2_coco/inference/resnet50_csv_50.h5
backbone = resnet50
class_names = [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29]
nms_threshold = 0.5
det_threshold = 0.8
max_detections = 300

[CAMERA]
width = 960 
height = 720 
K_test = [810.4968405 ,0.,487.55096072, 0., 810.61326022 ,354.6674888 ,  0.,   0.,  1.]
camPose = False

[ICP]
icp = True