{"id":38588,"date":"2025-05-20T17:43:53","date_gmt":"2025-05-20T09:43:53","guid":{"rendered":"https:\/\/www.wsisp.com\/helps\/38588.html"},"modified":"2025-05-20T17:43:53","modified_gmt":"2025-05-20T09:43:53","slug":"%e3%80%90%e6%9c%ba%e5%99%a8%e4%ba%ba%e3%80%91%e5%a4%8d%e7%8e%b0-3d-mem-%e5%85%b7%e8%ba%ab%e6%8e%a2%e7%b4%a2%e5%92%8c%e6%8e%a8%e7%90%86-3d%e5%9c%ba%e6%99%af%e8%ae%b0%e5%bf%86-cvpr-2025","status":"publish","type":"post","link":"https:\/\/www.wsisp.com\/helps\/38588.html","title":{"rendered":"\u3010\u673a\u5668\u4eba\u3011\u590d\u73b0 3D-Mem \u5177\u8eab\u63a2\u7d22\u548c\u63a8\u7406 | 3D\u573a\u666f\u8bb0\u5fc6 CVPR 2025"},"content":{"rendered":"

3D-Mem \u662f\u7528\u4e8e\u5177\u4f53\u63a2\u7d22<\/span>\u548c\u63a8\u7406<\/span>\u76843D\u573a\u666f\u8bb0\u5fc6<\/span>，\u6765\u81eaCVPR 2025.<\/p>\n

\u4f7f\u7528\u4fe1\u606f\u4e30\u5bcc\u7684\u591a\u89c6\u89d2\u56fe\u50cf，\u6765\u8868\u793a\u573a\u666f\u5e76\u6355\u6349\u5df2\u63a2\u7d22\u533a\u57df\u7684\u4e30\u5bcc\u89c6\u89c9\u4fe1\u606f，<\/span><\/p>\n

\u6574\u5408\u4e86\u57fa\u4e8e\u524d\u6cbf\u7684\u63a2\u7d22，\u4f7f\u667a\u80fd\u4f53\u80fd\u591f\u901a\u8fc7\u8003\u8651\u5df2\u77e5\u548c\u6f5c\u5728\u7684\u65b0\u4fe1\u606f，\u505a\u51fa\u660e\u667a\u7684\u51b3\u7b56\u3002<\/p>\n

\u672c\u6587\u5206\u4eab3D-Mem\u590d\u73b0\u548c\u6a21\u578b\u63a8\u7406\u7684\u8fc7\u7a0b～<\/span><\/p>\n

\u4e0b\u9762\u662f\u4e00\u4e2a\u8fd0\u884c\u793a\u4f8b\u7ed3\u679c：<\/p>\n

\"\"<\/p>\n

\u770b\u4e00\u4e0b\u5360\u7528\u5730\u56fe\u7684\u822a\u5411<\/p>\n

\"\"<\/p>\n

\u4e0b\u9762\u662f\u771f\u5b9e\u73af\u5883\u4e0b，\u5b98\u65b9\u8dd1\u7684demo，3D-Mem\u65e0\u9700\u8bad\u7ec3\u7684\u8bbe\u8ba1，\u53ef\u4ee5\u65e0\u7f1d\u9002\u5e94\u771f\u5b9e\u7684\u673a\u5668\u4eba，\u4ece\u800c\u5b9e\u73b0\u5728\u73b0\u5b9e\u4e16\u754c\u4e2d\u7684\u90e8\u7f72<\/p>\n

\"\"<\/p>\n

\u9879\u76ee\u5730\u5740：https:\/\/umass-embodied-agi.github.io\/3D-Mem\/<\/p>\n<\/p>\n

\u00a01\u3001\u521b\u5efaConda\u73af\u5883<\/h3>\n

\u9996\u5148\u521b\u5efa\u4e00\u4e2aConda\u73af\u5883，\u540d\u5b57\u4e3a3dmem，python\u7248\u672c\u4e3a3.9<\/p>\n

\u8fdb\u51653dmem\u73af\u5883<\/p>\n

conda create -n 3dmem python=3.9 -y
\nconda activate 3dmem <\/p>\n

\u7136\u540e\u4e0b\u8f7d\u4ee3\u7801，\u8fdb\u5165\u4ee3\u7801\u5de5\u7a0b：https:\/\/github.com\/UMass-Embodied-AGI\/3D-Mem<\/p>\n

git clone https:\/\/github.com\/UMass-Embodied-AGI\/3D-Mem.git
\ncd 3D-Mem <\/p>\n

2\u3001\u5b89\u88c5habitat\u6a21\u62df\u5668<\/h3>\n

\u6211\u9700\u8981\u5b89\u88c5habitat-sim==0.2.5\u3001headless \u548c\u00a0faiss-cpu<\/p>\n

conda install -c conda-forge -c aihabitat habitat-sim=0.2.5 headless faiss-cpu=1.7.4 -y <\/p>\n

\u7b49\u5f85\u5b89\u88c5\u5b8c\u6210～<\/p>\n<\/p>\n

3\u3001\u5b89\u88c5 torch \u548c pytorch3d<\/h3>\n

\u6267\u884c\u4e0b\u9762\u547d\u4ee4，\u8fdb\u884c\u5b89\u88c5torch：<\/p>\n

pip install torch==2.0.1 torchvision==0.15.2 –index-url https:\/\/download.pytorch.org\/whl\/cu118 <\/p>\n

\u518d\u5b89\u88c5pytorch3d：<\/p>\n

conda install https:\/\/anaconda.org\/pytorch3d\/pytorch3d\/0.7.4\/download\/linux-64\/pytorch3d-0.7.4-py39_cu118_pyt201.tar.bz2 -y <\/p>\n

4\u3001\u5b89\u88c5\u4f9d\u8d56\u5e93<\/h3>\n

\u6267\u884c\u4e0b\u9762\u547d\u4ee4\u8fdb\u884c\u5b89\u88c5：<\/p>\n

pip install omegaconf==2.3.0 open-clip-torch==2.26.1 ultralytics==8.2.31 supervision==0.21.0 opencv-python-headless==4.10.* \\\\
\n scikit-learn==1.4 scikit-image==0.22 open3d==0.18.0 hipart==1.0.4 openai==1.35.3 httpx==0.27.2 <\/p>\n

\u7b49\u5f85\u5b89\u88c5\u5b8c\u6210～<\/p>\n<\/p>\n

5\u3001\u5b89\u88c5clip<\/h3>\n

\u6267\u884c\u4e0b\u9762\u547d\u4ee4\u8fdb\u884c\u5b89\u88c5：<\/p>\n

pip install git+https:\/\/github.com\/openai\/CLIP.git <\/p>\n

\u6253\u5370\u4fe1\u606f<\/p>\n

Looking in indexes: https:\/\/mirrors.tuna.tsinghua.edu.cn\/pypi\/web\/simple
\nCollecting git+https:\/\/github.com\/openai\/CLIP.git
\n Cloning https:\/\/github.com\/openai\/CLIP.git to \/tmp\/pip-req-build-imrsh3kf
\n Running command git clone –filter=blob:none –quiet https:\/\/github.com\/openai\/CLIP.git \/tmp\/pip-req-build-imrsh3kf
\n Resolved https:\/\/github.com\/openai\/CLIP.git to commit dcba3cb2e2827b402d2701e7e1c7d9fed8a20ef1
\n Preparing metadata (setup.py) … done
\n…..
\nSuccessfully built clip
\nInstalling collected packages: clip
\nSuccessfully installed clip-1.0 <\/p>\n

clip\u7684\u4e3b\u8981\u601d\u8def\u6d41\u7a0b：<\/p>\n

\"\"<\/p>\n<\/p>\n

6\u3001\u4fee\u6539Hugging Face \u955c\u50cf\u6e90<\/h3>\n

\u4ee3\u7801\u4f1a\u81ea\u52a8\u4eceHugging Face\u4e0b\u8f7d\u6a21\u578b\u6743\u91cd，\u9700\u8981\u5148\u914d\u7f6e\u4e3a\u56fd\u5185\u7684\u955c\u50cf\u6e90<\/p>\n

\u00a0\u7f16\u8f91\u7528\u6237\u914d\u7f6e\u6587\u4ef6 ~\/.bashrc，\u8bbe\u7f6e\u4e3a export HF_ENDPOINT=https:\/\/hf-mirror.com<\/p>\n

\u6267\u884c\u4e0b\u9762\u547d\u4ee4：<\/p>\n

echo 'export HF_ENDPOINT=https:\/\/hf-mirror.com' >> ~\/.bashrc
\nsource ~\/.bashrc # \u7acb\u5373\u751f\u6548 <\/p>\n

\u9a8c\u8bc1\u73af\u5883\u53d8\u91cf\u200b\u200b，\u662f\u5426\u4fee\u6539\u6210\u529f：<\/p>\n

echo $HF_ENDPOINT <\/p>\n

\u6b63\u5e38\u4f1a\u8f93\u51fa：https:\/\/hf-mirror.com，\u8bf4\u660e\u8bbe\u7f6e\u6210\u529f\u5566～<\/p>\n<\/p>\n

7\u3001\u51c6\u5907HM3D\u6570\u636e\u96c6<\/h3>\n

\u6211\u4eec\u9700\u8981\u4e0b\u8f7d hm3d_v0.2<\/p>\n

\u4e0b\u8f7d\u5730\u5740：GitHub – matterport\/habitat-matterport-3dresearch<\/p>\n

\u9009\u62e9\u7684\u4e0b\u8f7d\u6587\u4ef6：hm3d-val-habitat-v0.2.tar<\/p>\n

\"\"<\/p>\n

\u7136\u540e\u653e\u5230data\u76ee\u5f55\u4e0b：<\/p>\n

\"\"<\/p>\n<\/p>\n

8\u3001\u51c6\u5907gpt-4o\u7684Api<\/h3>\n

\u63a8\u8350\u4f7f\u7528\u56fd\u5185\u7684\u4f9b\u5e94\u5546，\u6bd4\u8f83\u7a33\u5b9a：https:\/\/ai.nengyongai.cn\/register?aff=RQt3<\/p>\n

\u9996\u5148\u201c\u6dfb\u52a0\u4ee4\u724c\u201d，\u8bbe\u7f6e\u989d\u5ea6（\u6bd4\u59825\u5757\u94b1），\u70b9\u51fb\u67e5\u770b\u5c31\u80fd\u770b\u5230Key\u5566<\/p>\n

\"\"<\/p>\n

\u00a0\u7136\u540e\u586b\u5199\u5230 src\/const.py\u4e2d<\/p>\n

# about habitat scene
\nINVALID_SCENE_ID = []<\/p>\n

# about chatgpt api
\nEND_POINT = "https:\/\/ai.nengyongai.cn\/v1"
\nOPENAI_KEY = "xxxxxxxxxxxxxxxxxxxxx"<\/p>\n

\u70b9\u51fb\u6a21\u578b\u5217\u8868，\u80fd\u67e5\u770b\u652f\u6301\u7684\u6a21\u578b：<\/p>\n

\"\"<\/p>\n

\u770b\u4e00\u4e0b\u4f7f\u7528\u60c5\u51b5：<\/p>\n

\"\" \u00a0<\/p>\n<\/p>\n

9\u3001\u8fd0\u884c\u6a21\u578b\u63a8\u7406<\/h3>\n

\u67e5\u770b\u914d\u7f6e\u6587\u4ef6 cfg\/eval_aeqa.yaml<\/p>\n

# \u901a\u7528\u8bbe\u7f6e
\nseed: 77 # \u968f\u673a\u79cd\u5b50
\nexp_name: "exp_eval_aeqa" # \u5b9e\u9a8c\u540d\u79f0
\noutput_parent_dir: "results" # \u8f93\u51fa\u6587\u4ef6\u5939\u7684\u7236\u76ee\u5f55
\nscene_dataset_config_path: "data\/hm3d_annotated_basis.scene_dataset_config.json" # \u573a\u666f\u6570\u636e\u96c6\u914d\u7f6e\u6587\u4ef6\u8def\u5f84
\nscene_data_path: "data\/hm3d_v0.2\/" # \u573a\u666f\u6570\u636e\u8def\u5f84
\nquestions_list_path: 'data\/aeqa_questions-41.json' # \u95ee\u9898\u5217\u8868\u6587\u4ef6\u8def\u5f84<\/p>\n

concept_graph_config_path: "cfg\/concept_graph_default.yaml" # \u6982\u5ff5\u56fe\u914d\u7f6e\u6587\u4ef6\u8def\u5f84<\/p>\n

# \u4e3b\u8981\u8bbe\u7f6e
\nchoose_every_step: true # \u662f\u5426\u5728\u6bcf\u4e00\u6b65\u90fd\u67e5\u8be2\u89c6\u89c9\u8bed\u8a00\u6a21\u578b（VLM），\u8fd8\u662f\u4ec5\u5728\u5230\u8fbe\u5bfc\u822a\u76ee\u6807\u540e\u67e5\u8be2
\negocentric_views: true # \u662f\u5426\u5728\u63d0\u793a\u89c6\u89c9\u8bed\u8a00\u6a21\u578b\u65f6\u6dfb\u52a0\u81ea\u6211\u4e2d\u5fc3\u89c6\u89d2
\nprefiltering: true # \u662f\u5426\u4f7f\u7528\u9884\u7b5b\u9009（\u5b9e\u9645\u4e0a\u4e0d\u80fd\u5173\u95ed，\u5426\u5219\u4f1a\u8d85\u51fa\u4e0a\u4e0b\u6587\u957f\u5ea6\u9650\u5236）
\ntop_k_categories: 10 # \u5728\u9884\u7b5b\u9009\u8fc7\u7a0b\u4e2d\u4fdd\u7559\u4e0e\u76ee\u6807\u6700\u76f8\u5173\u7684\u524d k \u4e2a\u7c7b\u522b<\/p>\n

# \u5173\u4e8e\u68c0\u6d4b\u6a21\u578b
\nyolo_model_name: yolov8x-world.pt # YOLO \u6a21\u578b\u540d\u79f0
\nsam_model_name: sam_l.pt # SAM \u6a21\u578b\u540d\u79f0
\nclass_set: scannet200 # \u4f7f\u7528 200 \u7c7b\u522b\u7684\u6570\u636e\u96c6\u7528\u4e8e YOLO-world \u68c0\u6d4b\u5668<\/p>\n

# \u5173\u4e8e\u5feb\u7167\u805a\u7c7b
\nmin_detection: 1 # \u6700\u5c0f\u68c0\u6d4b\u6570\u91cf<\/p>\n

# \u76f8\u673a\u548c\u56fe\u50cf\u8bbe\u7f6e
\ncamera_height: 1.5 # \u76f8\u673a\u9ad8\u5ea6（\u5355\u4f4d：\u7c73）
\ncamera_tilt_deg: -30 # \u76f8\u673a\u503e\u659c\u89d2\u5ea6（\u5355\u4f4d：\u5ea6）
\nimg_width: 1280 # \u56fe\u50cf\u5bbd\u5ea6（\u5355\u4f4d：\u50cf\u7d20）
\nimg_height: 1280 # \u56fe\u50cf\u9ad8\u5ea6（\u5355\u4f4d：\u50cf\u7d20）
\nhfov: 120 # \u6c34\u5e73\u89c6\u573a\u89d2（\u5355\u4f4d：\u5ea6）<\/p>\n

# \u662f\u5426\u4fdd\u5b58\u53ef\u89c6\u5316\u7ed3\u679c（\u8fd9\u4f1a\u6bd4\u8f83\u6162）
\nsave_visualization: true<\/p>\n

# \u7528\u4e8e\u63d0\u793a GPT-4O \u7684\u56fe\u50cf\u5927\u5c0f
\nprompt_h: 360 # \u63d0\u793a\u56fe\u50cf\u9ad8\u5ea6（\u5355\u4f4d：\u50cf\u7d20）
\nprompt_w: 360 # \u63d0\u793a\u56fe\u50cf\u5bbd\u5ea6（\u5355\u4f4d：\u50cf\u7d20）<\/p>\n

# \u5bfc\u822a\u8bbe\u7f6e
\nnum_step: 50 # \u6700\u5927\u5bfc\u822a\u6b65\u6570
\ninit_clearance: 0.3 # \u521d\u59cb\u907f\u78b0\u8ddd\u79bb（\u5355\u4f4d：\u7c73）
\nextra_view_phase_1: 2 # \u7b2c\u4e00\u9636\u6bb5\u989d\u5916\u89c6\u89d2\u7684\u6570\u91cf
\nextra_view_angle_deg_phase_1: 60 # \u7b2c\u4e00\u9636\u6bb5\u6bcf\u4e2a\u989d\u5916\u89c6\u89d2\u4e4b\u95f4\u7684\u89d2\u5ea6（\u5355\u4f4d：\u5ea6）
\nextra_view_phase_2: 6 # \u7b2c\u4e8c\u9636\u6bb5\u989d\u5916\u89c6\u89d2\u7684\u6570\u91cf
\nextra_view_angle_deg_phase_2: 40 # \u7b2c\u4e8c\u9636\u6bb5\u6bcf\u4e2a\u989d\u5916\u89c6\u89d2\u4e4b\u95f4\u7684\u89d2\u5ea6（\u5355\u4f4d：\u5ea6）<\/p>\n

# \u5173\u4e8e TSDF\u3001\u6df1\u5ea6\u56fe\u548c\u8fb9\u754c\u66f4\u65b0
\nexplored_depth: 1.7 # \u5df2\u63a2\u7d22\u6df1\u5ea6（\u5355\u4f4d：\u7c73）
\ntsdf_grid_size: 0.1 # TSDF \u7f51\u683c\u5927\u5c0f（\u5355\u4f4d：\u7c73）
\nmargin_w_ratio: 0.25 # \u5bbd\u5ea6\u65b9\u5411\u7684\u8fb9\u754c\u6bd4\u4f8b
\nmargin_h_ratio: 0.6 # \u9ad8\u5ea6\u65b9\u5411\u7684\u8fb9\u754c\u6bd4\u4f8b
\nplanner: # \u89c4\u5212\u5668\u8bbe\u7f6e
\n eps: 1 # \u89c4\u5212\u5668\u7684\u7cbe\u5ea6
\n max_dist_from_cur_phase_1: 1 # \u7b2c\u4e00\u9636\u6bb5\u672a\u627e\u5230\u76ee\u6807\u65f6，\u63a2\u7d22\u8fb9\u754c\u7684\u6b65\u957f（\u5355\u4f4d：\u7c73）
\n max_dist_from_cur_phase_2: 1 # \u7b2c\u4e8c\u9636\u6bb5\u627e\u5230\u76ee\u6807\u540e，\u63a5\u8fd1\u76ee\u6807\u7684\u6b65\u957f（\u5355\u4f4d：\u7c73）
\n final_observe_distance: 0.75 # \u7b2c\u4e8c\u9636\u6bb5\u627e\u5230\u4e00\u4e2a\u8ddd\u79bb\u76ee\u6807\u5bf9\u8c61\u6b64\u8ddd\u79bb\u7684\u5730\u65b9\u8fdb\u884c\u89c2\u5bdf（\u5355\u4f4d：\u7c73）
\n surrounding_explored_radius: 0.7 # \u5468\u56f4\u5df2\u63a2\u7d22\u533a\u57df\u7684\u534a\u5f84（\u5355\u4f4d：\u7c73）<\/p>\n

# \u5173\u4e8e\u8fb9\u754c\u9009\u62e9
\n frontier_edge_area_min: 4 # \u8fb9\u754c\u8fb9\u7f18\u6700\u5c0f\u9762\u79ef
\n frontier_edge_area_max: 6 # \u8fb9\u754c\u8fb9\u7f18\u6700\u5927\u9762\u79ef
\n frontier_area_min: 8 # \u8fb9\u754c\u6700\u5c0f\u9762\u79ef
\n frontier_area_max: 9 # \u8fb9\u754c\u6700\u5927\u9762\u79ef
\n min_frontier_area: 20 # \u8fb9\u754c\u81f3\u5c11\u9700\u8981\u7684\u50cf\u7d20\u6570\u91cf
\n max_frontier_angle_range_deg: 150 # \u8fb9\u754c\u4e2d\u50cf\u7d20\u6240\u5f20\u89d2\u5ea6\u7684\u6700\u5927\u503c（\u5355\u4f4d：\u5ea6）
\n region_equal_threshold: 0.95 # \u533a\u57df\u76f8\u7b49\u7684\u9608\u503c<\/p>\n

# \u5173\u4e8e\u573a\u666f\u56fe\u6784\u5efa
\nscene_graph:
\n confidence: 0.003 # \u7f6e\u4fe1\u5ea6\u9608\u503c
\n nms_threshold: 0.1 # \u975e\u6781\u5927\u503c\u6291\u5236\u9608\u503c
\n iou_threshold: 0.5 # \u4ea4\u5e76\u6bd4\u9608\u503c
\n obj_include_dist: 3.5 # \u5305\u542b\u76ee\u6807\u5bf9\u8c61\u7684\u8ddd\u79bb（\u5355\u4f4d：\u7c73）
\n target_obj_iou_threshold: 0.6 # \u76ee\u6807\u5bf9\u8c61\u7684\u4ea4\u5e76\u6bd4\u9608\u503c <\/p>\n

\u8fd0\u884c\u4e0b\u9762\u4ee3\u7801，\u751f\u6210 A-EQA \u6570\u636e\u96c6\u7684\u9884\u6d4b\u7ed3\u679c<\/span>：<\/p>\n

python run_aeqa_evaluation.py -cf cfg\/eval_aeqa.yaml <\/p>\n

\u8fd0\u884c\u7a0b\u5e8f\u540e，\u4f1a\u8054\u7f51\u4e0b\u8f7d\u4e00\u4e9b\u6a21\u578b\u6743\u91cd，<\/p>\n

\u5305\u62ec：yolov8x-world.pt\u3001sam_l.pt\u3001open_clip_pytorch_model.bin\u3001ViT-B-32.pt\u7b49<\/p>\n

\u4e0b\u9762\u662f\u8fd0\u884c\u4fe1\u606f：<\/p>\n

00:00:00 – ***** Running exp_eval_aeqa *****
\n00:00:00 – Total number of questions: 41
\n00:00:00 – number of questions after splitting: 41
\n00:00:00 – question path: data\/aeqa_questions-41.json
\nDownloading https:\/\/github.com\/ultralytics\/assets\/releases\/download\/v8.2.0\/yolov8x-world.pt to 'yolov8x-world.pt'…
\n100%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588| 141M\/141M [04:04<00:00, 605kB\/s]
\n00:04:09 – Load YOLO model yolov8x-world.pt successful!
\nDownloading https:\/\/github.com\/ultralytics\/assets\/releases\/download\/v8.2.0\/sam_l.pt to 'sam_l.pt'…
\n100%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588| 1.16G\/1.16G [11:56<00:00, 1.74MB\/s]
\n00:16:12 – Load SAM model sam_l.pt successful!
\n00:16:12 – Loaded ViT-B-32 model config.
\nopen_clip_pytorch_model.bin: 70%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258e | 440M\/626M [03:17<01:11, 2.58MB\/s]<\/p>\n

…. <\/p>\n

\u5f53\u4e0b\u8f7d\u548c\u52a0\u8f7d\u6210\u529f\u540e，\u4f1a\u663e\u793a：<\/p>\n

00:00:00 – ***** Running exp_eval_aeqa *****
\n00:00:00 – Total number of questions: 41
\n00:00:00 – number of questions after splitting: 41
\n00:00:00 – question path: data\/aeqa_questions-41.json
\n00:00:00 – Load YOLO model yolov8x-world.pt successful!
\n00:00:02 – Load SAM model sam_l.pt successful!
\n00:00:02 – Loaded ViT-B-32 model config.
\n00:00:04 – Loading pretrained ViT-B-32 weights (laion2b_s34b_b79k).
\n00:00:05 – Load CLIP model successful!
\n00:00:05 – Question 00c2be2a-1377-4fae-a889-30936b7890c3 already processed
\n00:00:05 – Question 013bb857-f47d-4b50-add4-023cc4ff414c already processed
\n00:00:05 –
\n========
\nIndex: 2 Scene: 00848-ziup5kvtCCR
\n00:00:05 – semantic_texture_path: data\/hm3d_v0.2\/val\/00848-ziup5kvtCCR\/ziup5kvtCCR.semantic.glb or scene_semantic_annotation_path: data\/hm3d_v0.2\/val\/00848-ziup5kvtCCR\/ziup5kvtCCR.semantic.txt does not exist
\n00:00:06 – Loaded 192 classes from scannet 200: data\/scannet200_classes.txt!!!
\n00:00:06 – Load scene 00848-ziup5kvtCCR successfully without semantic texture
\n00:00:10 – <\/p>\n

Question id 01fcc568-f51e-4e12-b976-5dc8d554135a initialization successful!
\n00:00:10 –
\n== step: 0
\n00:00:11 – Done! Execution time of detections_to_obj_pcd_and_bbox function: 0.12 seconds
\n00:00:13 – Done! Execution time of detections_to_obj_pcd_and_bbox function: 0.09 seconds
\n00:00:15 – Done! Execution time of detections_to_obj_pcd_and_bbox function: 0.08 seconds
\n00:00:16 – Done! Execution time of detections_to_obj_pcd_and_bbox function: 0.05 seconds
\n00:00:17 – Done! Execution time of detections_to_obj_pcd_and_bbox function: 0.04 seconds
\n00:00:18 – Done! Execution time of detections_to_obj_pcd_and_bbox function: 0.05 seconds
\n00:00:19 – Done! Execution time of detections_to_obj_pcd_and_bbox function: 0.07 seconds
\n00:00:20 – Step 0, update snapshots, 25 objects, 6 snapshots
\n00:00:23 – HTTP Request: POST https:\/\/ai.nengyongai.cn\/v1\/chat\/completions "HTTP\/1.1 200 OK"
\n00:00:23 – Prefiltering selected classes: ['sofa chair', 'couch', 'pillow', 'coffee table', 'cabinet']
\n00:00:23 – Prefiltering snapshot: 6 -> 3
\n00:00:23 – Input prompt:
\n00:00:23 – Task: You are an agent in an indoor scene tasked with answering questions by observing the surroundings and exploring the environment. To answer the question, you are required to choose either a Snapshot as the answer or a Frontier to further explore.
\nDefinitions:
\nSnapshot: A focused observation of several objects. Choosing a Snapshot means that this snapshot image contains enough information for you to answer the question. If you choose a Snapshot, you need to directly give an answer to the question. If you don't have enough information to give an answer, then don't choose a Snapshot.
\nFrontier: An observation of an unexplored region that could potentially lead to new information for answering the question. Selecting a frontier means that you will further explore that direction. If you choose a Frontier, you need to explain why you would like to choose that direction to explore.
\nQuestion: Where is the teddy bear?
\nSelect the Frontier\/Snapshot that would help find the answer of the question.
\nThe following is the egocentric view of the agent in forward direction: [iVBORw0KGg…]
\nThe followings are all the snapshots that you can choose (followed with contained object classes)
\nPlease note that the contained classes may not be accurate (wrong classes\/missing classes) due to the limitation of the object detection model. So you still need to utilize the images to make decisions.
\nSnapshot 0 [iVBORw0KGg…]coffee table, couch, pillow
\nSnapshot 1 [iVBORw0KGg…]coffee table, pillow, sofa chair
\nSnapshot 2 [iVBORw0KGg…]cabinet, couch
\nThe followings are all the Frontiers that you can explore:
\nFrontier 0 [iVBORw0KGg…]
\nFrontier 1 [iVBORw0KGg…]
\nPlease provide your answer in the following format: 'Snapshot i
\n[Answer]' or 'Frontier i
\n[Reason]', where i is the index of the snapshot or frontier you choose. For example, if you choose the first snapshot, you can return 'Snapshot 0
\nThe fruit bowl is on the kitchen counter.'. If you choose the second frontier, you can return 'Frontier 1
\nI see a door that may lead to the living room.'.
\nNote that if you choose a snapshot to answer the question, (1) you should give a direct answer that can be understood by others. Don't mention words like 'snapshot', 'on the left of the image', etc; (2) you can also utilize other snapshots, frontiers and egocentric views to gather more information, but you should always choose one most relevant snapshot to answer the question.<\/p>\n

00:00:32 – HTTP Request: POST https:\/\/ai.nengyongai.cn\/v1\/chat\/completions "HTTP\/1.1 200 OK"
\n00:00:32 – Response: [frontier 0]
\nReason: [I would like to explore the hallway further as it may lead to other rooms where the teddy bear might be located.]
\n00:00:32 – Prediction: frontier, 0
\n00:00:32 – Next choice: Frontier at [79 33]
\nUserWarning: *c* argument looks like a single numeric RGB or RGBA sequence, which should be avoided as value-mapping will have precedence in case its length matches with *x* & *y*. Please use the *color* keyword-argument or provide a 2D array with a single row if you intend to specify the same RGB or RGBA value for all points.
\n00:00:33 – Current position: [ 0.11692 0.021223 6.1057], 1.005
\n00:00:34 –
\n== step: 1<\/p>\n

\u53ef\u89c6\u5316\u7684\u7ed3\u679c\u4fdd\u5b58\u5728：results\/exp_eval_aeqa \u4e2d<\/p>\n

\"\"<\/p>\n

\u770b\u4e00\u4e0b\u5360\u7528\u5730\u56fe，\u89c4\u5212\u822a\u5411（1）<\/p>\n

\"\"<\/p>\n

\u89c4\u5212\u822a\u5411（2）<\/p>\n

\"\"<\/p>\n

\u89c4\u5212\u822a\u5411（3）<\/p>\n

\"\"<\/p>\n<\/p>\n

\u6a21\u578b\u63a8\u7406\u793a\u4f8b2<\/span><\/p>\n

\u5bf9\u5e94\u7684\u914d\u7f6e\u6587\u4ef6\u662f：cfg\/eval_goatbench.yaml<\/p>\n

\u8fd0\u884c\u4ee3\u7801，\u751f\u6210 GOAT-Bench \u6570\u636e\u96c6\u7684\u9884\u6d4b\u7ed3\u679c：<\/p>\n

python run_goatbench_evaluation.py -cf cfg\/eval_goatbench.yaml <\/p>\n

GOAT-Bench \u4e3a\u6bcf\u4e2a\u573a\u666f\u63d0\u4f9b\u4e86 10 \u4e2a\u63a2\u7d22\u60c5\u8282，\u5e76\u4e14\u7531\u4e8e\u65f6\u95f4\u548c\u8d44\u6e90\u7684\u9650\u5236，\u9ed8\u8ba4\u53ea\u6d4b\u8bd5\u7b2c\u4e00\u60c5\u8282\u3002\u00a0<\/span><\/p>\n

\u6211\u4eec\u8fd8\u53ef\u4ee5\u901a\u8fc7\u8bbe\u7f6e\u6765\u6307\u5b9a\u8981\u8bc4\u4f30\u6bcf\u4e2a\u573a\u666f\u7684\u60c5\u8282 –split\u3002<\/span><\/p>\n<\/p>\n

\u5206\u4eab\u5b8c\u6210～<\/p>\n

\u00a0\u76f8\u5173\u6587\u7ae0\u63a8\u8350：<\/p>\n

UniGoal \u5177\u8eab\u5bfc\u822a | \u901a\u7528\u96f6\u6837\u672c\u76ee\u6807\u5bfc\u822a CVPR 2025-CSDN\u535a\u5ba2<\/p>\n

\u3010\u673a\u5668\u4eba\u3011\u590d\u73b0 UniGoal \u5177\u8eab\u5bfc\u822a | \u901a\u7528\u96f6\u6837\u672c\u76ee\u6807\u5bfc\u822a CVPR 2025-CSDN\u535a\u5ba2<\/p>\n

\u3010\u673a\u5668\u4eba\u3011\u590d\u73b0 ECoT \u5177\u8eab\u601d\u7ef4\u94fe\u63a8\u7406-CSDN\u535a\u5ba2<\/p>\n

\u3010\u673a\u5668\u4eba\u3011\u590d\u73b0 SG-Nav \u5177\u8eab\u5bfc\u822a | \u96f6\u6837\u672c\u5bf9\u8c61\u5bfc\u822a\u7684 \u5728\u7ebf3D\u573a\u666f\u56fe\u63d0\u793a-CSDN\u535a\u5ba2<\/p>\n

\u3010\u673a\u5668\u4eba\u3011\u590d\u73b0 WMNav \u5177\u8eab\u5bfc\u822a | \u5c06VLM\u96c6\u6210\u5230\u4e16\u754c\u6a21\u578b\u4e2d-CSDN\u535a\u5ba2<\/p><\/p>\n","protected":false},"excerpt":{"rendered":"

\u6587\u7ae0\u6d4f\u89c8\u9605\u8bfb538\u6b21\uff0c\u70b9\u8d5e9\u6b21\uff0c\u6536\u85cf6\u6b21\u3002\u672c\u6587\u8be6\u7ec6\u4ecb\u7ecd\u4e863D-Mem\u6a21\u578b\u7684\u590d\u73b0\u548c\u63a8\u7406\u8fc7\u7a0b\u3002\u9996\u5148\uff0c\u521b\u5efa\u4e86\u4e00\u4e2a\u540d\u4e3a3dmem\u7684Conda\u73af\u5883\uff0c\u5e76\u5b89\u88c5\u4e86\u5fc5\u8981\u7684\u4f9d\u8d56\u5e93\uff0c\u5982habitat-sim\u3001faiss-cpu\u3001torch\u548cpytorch3d\u3002\u63a5\u7740\uff0c\u914d\u7f6e\u4e86HuggingFace\u7684\u56fd\u5185\u955c\u50cf\u6e90\uff0c\u5e76\u4e0b\u8f7d\u4e86HM3D\u6570\u636e\u96c6\u3002\u7136\u540e\uff0c\u901a\u8fc7\u8fd0\u884c\u4ee3\u7801\u751f\u6210A-EQA\u6570\u636e\u96c6\u7684\u9884\u6d4b\u7ed3\u679c\uff0c\u8fc7\u7a0b\u4e2d\u4f1a\u4e0b\u8f7d\u5e76\u52a0\u8f7d\u591a\u4e2a\u6a21\u578b\u6743\u91cd\uff0c\u5982yolov8x-world.pt\u548csam_l.pt\u3002\u6700\u540e\uff0c\u5c55\u793a\u4e86\u6a21\u578b\u63a8\u7406\u7684\u53ef\u89c6\u5316\u7ed3\u679c\uff0c\u5e76\u63d0\u4f9b\u4e86\u751f\u6210GOAT-Bench\u6570\u636e\u96c6\u9884\u6d4b\u7ed3\u679c\u7684\u4ee3\u7801\u3002\u6574\u4e2a\u8fc7\u7a0b\u6db5\u76d6\u4e86\u73af\u5883\u642d<\/p>\n","protected":false},"author":2,"featured_media":38575,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[3410,3412,3409,3411,384],"topic":[],"class_list":["post-38588","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-server","tag-3d-mem","tag-3d","tag-3409","tag-3411","tag-384"],"yoast_head":"\n\u3010\u673a\u5668\u4eba\u3011\u590d\u73b0 3D-Mem \u5177\u8eab\u63a2\u7d22\u548c\u63a8\u7406 | 3D\u573a\u666f\u8bb0\u5fc6 CVPR 2025 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.wsisp.com\/helps\/38588.html\" \/>\n<meta property=\"og:locale\" content=\"zh_CN\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"\u3010\u673a\u5668\u4eba\u3011\u590d\u73b0 3D-Mem \u5177\u8eab\u63a2\u7d22\u548c\u63a8\u7406 | 3D\u573a\u666f\u8bb0\u5fc6 CVPR 2025 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"og:description\" content=\"\u6587\u7ae0\u6d4f\u89c8\u9605\u8bfb538\u6b21\uff0c\u70b9\u8d5e9\u6b21\uff0c\u6536\u85cf6\u6b21\u3002\u672c\u6587\u8be6\u7ec6\u4ecb\u7ecd\u4e863D-Mem\u6a21\u578b\u7684\u590d\u73b0\u548c\u63a8\u7406\u8fc7\u7a0b\u3002\u9996\u5148\uff0c\u521b\u5efa\u4e86\u4e00\u4e2a\u540d\u4e3a3dmem\u7684Conda\u73af\u5883\uff0c\u5e76\u5b89\u88c5\u4e86\u5fc5\u8981\u7684\u4f9d\u8d56\u5e93\uff0c\u5982habitat-sim\u3001faiss-cpu\u3001torch\u548cpytorch3d\u3002\u63a5\u7740\uff0c\u914d\u7f6e\u4e86HuggingFace\u7684\u56fd\u5185\u955c\u50cf\u6e90\uff0c\u5e76\u4e0b\u8f7d\u4e86HM3D\u6570\u636e\u96c6\u3002\u7136\u540e\uff0c\u901a\u8fc7\u8fd0\u884c\u4ee3\u7801\u751f\u6210A-EQA\u6570\u636e\u96c6\u7684\u9884\u6d4b\u7ed3\u679c\uff0c\u8fc7\u7a0b\u4e2d\u4f1a\u4e0b\u8f7d\u5e76\u52a0\u8f7d\u591a\u4e2a\u6a21\u578b\u6743\u91cd\uff0c\u5982yolov8x-world.pt\u548csam_l.pt\u3002\u6700\u540e\uff0c\u5c55\u793a\u4e86\u6a21\u578b\u63a8\u7406\u7684\u53ef\u89c6\u5316\u7ed3\u679c\uff0c\u5e76\u63d0\u4f9b\u4e86\u751f\u6210GOAT-Bench\u6570\u636e\u96c6\u9884\u6d4b\u7ed3\u679c\u7684\u4ee3\u7801\u3002\u6574\u4e2a\u8fc7\u7a0b\u6db5\u76d6\u4e86\u73af\u5883\u642d\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.wsisp.com\/helps\/38588.html\" \/>\n<meta property=\"og:site_name\" content=\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"article:published_time\" content=\"2025-05-20T09:43:53+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.wsisp.com\/helps\/wp-content\/uploads\/2025\/05\/20250520094331-682c4ec3c52a4.png\" \/>\n<meta name=\"author\" content=\"admin\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u4f5c\u8005\" \/>\n\t<meta name=\"twitter:data1\" content=\"admin\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4\" \/>\n\t<meta name=\"twitter:data2\" content=\"8 \u5206\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/38588.html\",\"url\":\"https:\/\/www.wsisp.com\/helps\/38588.html\",\"name\":\"\u3010\u673a\u5668\u4eba\u3011\u590d\u73b0 3D-Mem \u5177\u8eab\u63a2\u7d22\u548c\u63a8\u7406 | 3D\u573a\u666f\u8bb0\u5fc6 CVPR 2025 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"isPartOf\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\"},\"datePublished\":\"2025-05-20T09:43:53+00:00\",\"dateModified\":\"2025-05-20T09:43:53+00:00\",\"author\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\"},\"breadcrumb\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/38588.html#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.wsisp.com\/helps\/38588.html\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/38588.html#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/www.wsisp.com\/helps\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u3010\u673a\u5668\u4eba\u3011\u590d\u73b0 3D-Mem \u5177\u8eab\u63a2\u7d22\u548c\u63a8\u7406 | 3D\u573a\u666f\u8bb0\u5fc6 CVPR 2025\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\",\"url\":\"https:\/\/www.wsisp.com\/helps\/\",\"name\":\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"description\":\"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\",\"name\":\"admin\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"contentUrl\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"caption\":\"admin\"},\"sameAs\":[\"http:\/\/wp.wsisp.com\"],\"url\":\"https:\/\/www.wsisp.com\/helps\/author\/admin\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"\u3010\u673a\u5668\u4eba\u3011\u590d\u73b0 3D-Mem \u5177\u8eab\u63a2\u7d22\u548c\u63a8\u7406 | 3D\u573a\u666f\u8bb0\u5fc6 CVPR 2025 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.wsisp.com\/helps\/38588.html","og_locale":"zh_CN","og_type":"article","og_title":"\u3010\u673a\u5668\u4eba\u3011\u590d\u73b0 3D-Mem \u5177\u8eab\u63a2\u7d22\u548c\u63a8\u7406 | 3D\u573a\u666f\u8bb0\u5fc6 CVPR 2025 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","og_description":"\u6587\u7ae0\u6d4f\u89c8\u9605\u8bfb538\u6b21\uff0c\u70b9\u8d5e9\u6b21\uff0c\u6536\u85cf6\u6b21\u3002\u672c\u6587\u8be6\u7ec6\u4ecb\u7ecd\u4e863D-Mem\u6a21\u578b\u7684\u590d\u73b0\u548c\u63a8\u7406\u8fc7\u7a0b\u3002\u9996\u5148\uff0c\u521b\u5efa\u4e86\u4e00\u4e2a\u540d\u4e3a3dmem\u7684Conda\u73af\u5883\uff0c\u5e76\u5b89\u88c5\u4e86\u5fc5\u8981\u7684\u4f9d\u8d56\u5e93\uff0c\u5982habitat-sim\u3001faiss-cpu\u3001torch\u548cpytorch3d\u3002\u63a5\u7740\uff0c\u914d\u7f6e\u4e86HuggingFace\u7684\u56fd\u5185\u955c\u50cf\u6e90\uff0c\u5e76\u4e0b\u8f7d\u4e86HM3D\u6570\u636e\u96c6\u3002\u7136\u540e\uff0c\u901a\u8fc7\u8fd0\u884c\u4ee3\u7801\u751f\u6210A-EQA\u6570\u636e\u96c6\u7684\u9884\u6d4b\u7ed3\u679c\uff0c\u8fc7\u7a0b\u4e2d\u4f1a\u4e0b\u8f7d\u5e76\u52a0\u8f7d\u591a\u4e2a\u6a21\u578b\u6743\u91cd\uff0c\u5982yolov8x-world.pt\u548csam_l.pt\u3002\u6700\u540e\uff0c\u5c55\u793a\u4e86\u6a21\u578b\u63a8\u7406\u7684\u53ef\u89c6\u5316\u7ed3\u679c\uff0c\u5e76\u63d0\u4f9b\u4e86\u751f\u6210GOAT-Bench\u6570\u636e\u96c6\u9884\u6d4b\u7ed3\u679c\u7684\u4ee3\u7801\u3002\u6574\u4e2a\u8fc7\u7a0b\u6db5\u76d6\u4e86\u73af\u5883\u642d","og_url":"https:\/\/www.wsisp.com\/helps\/38588.html","og_site_name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","article_published_time":"2025-05-20T09:43:53+00:00","og_image":[{"url":"https:\/\/www.wsisp.com\/helps\/wp-content\/uploads\/2025\/05\/20250520094331-682c4ec3c52a4.png"}],"author":"admin","twitter_card":"summary_large_image","twitter_misc":{"\u4f5c\u8005":"admin","\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4":"8 \u5206"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.wsisp.com\/helps\/38588.html","url":"https:\/\/www.wsisp.com\/helps\/38588.html","name":"\u3010\u673a\u5668\u4eba\u3011\u590d\u73b0 3D-Mem \u5177\u8eab\u63a2\u7d22\u548c\u63a8\u7406 | 3D\u573a\u666f\u8bb0\u5fc6 CVPR 2025 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","isPartOf":{"@id":"https:\/\/www.wsisp.com\/helps\/#website"},"datePublished":"2025-05-20T09:43:53+00:00","dateModified":"2025-05-20T09:43:53+00:00","author":{"@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41"},"breadcrumb":{"@id":"https:\/\/www.wsisp.com\/helps\/38588.html#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.wsisp.com\/helps\/38588.html"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.wsisp.com\/helps\/38588.html#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/www.wsisp.com\/helps"},{"@type":"ListItem","position":2,"name":"\u3010\u673a\u5668\u4eba\u3011\u590d\u73b0 3D-Mem \u5177\u8eab\u63a2\u7d22\u548c\u63a8\u7406 | 3D\u573a\u666f\u8bb0\u5fc6 CVPR 2025"}]},{"@type":"WebSite","@id":"https:\/\/www.wsisp.com\/helps\/#website","url":"https:\/\/www.wsisp.com\/helps\/","name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","description":"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41","name":"admin","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/","url":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","contentUrl":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","caption":"admin"},"sameAs":["http:\/\/wp.wsisp.com"],"url":"https:\/\/www.wsisp.com\/helps\/author\/admin"}]}},"_links":{"self":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/38588","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/comments?post=38588"}],"version-history":[{"count":0,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/38588\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/media\/38575"}],"wp:attachment":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/media?parent=38588"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/categories?post=38588"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/tags?post=38588"},{"taxonomy":"topic","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/topic?post=38588"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}