Merge branch 'dev' into low-hanging-fruits

.github/workflows/conda-test.yml

@@ -52,7 +52,7 @@ jobs:
         if: steps.cache.outputs.cache-hit != 'true'
 
       - name: Install happypose
-        run: pip install -e ".[evaluation]"
+        run: pip install -e .
 
       - name: Download pre-trained models required for tests
         run: |

.github/workflows/pip-test.yml

@@ -30,7 +30,7 @@ jobs:
         run: pip install -U pip
 
       - name: Install happypose
-        run: pip install ".[cpu,evaluation,pypi]" --extra-index-url https://download.pytorch.org/whl/cpu
+        run: pip install ".[cpu,pypi]" --extra-index-url https://download.pytorch.org/whl/cpu
 
       - name: Download pre-trained models required for tests
         run: |

.github/workflows/poetry-test.yml

@@ -31,7 +31,7 @@ jobs:
           cache: poetry
 
       - name: Install happypose
-        run: poetry install --with dev -E cpu -E evaluation -E pypi
+        run: poetry install --with dev -E cpu -E pypi
 
       - name: Download pre-trained models required for tests
         run: |

README.md

@@ -46,7 +46,6 @@ pip install .[pypi,cpu] --extra-index-url https://download.pytorch.org/whl/cpu
 
 - `cpu`: required to get pytorch CPU from PyPI (don't use this for GPU or with conda)
 - `gpu`: required to get pytorch GPU from PyPI (don't use this for CPU or with conda)
-- `evaluation`: installs bop_toolkit
 - `multiview`: installs cosypose c++ extension
 - `pypi`: install pinocchio & opencv from PyPI (don't use this with conda)
 

docs/book/megapose/download_data.md

@@ -19,23 +19,11 @@ Megapose can use pretrained detectors from CosyPose, which can be downloaded to
 
 ```sh
 # hope
-python -m happypose.toolbox.utils.download --cosypose_models \
-          detector-bop-hope-pbr--15246 \
-          coarse-bop-hope-pbr--225203 \
-          refiner-bop-hope-pbr--955392
-
+python -m happypose.toolbox.utils.download --cosypose_models detector-bop-hope-pbr--15246
 # ycbv
-
-python -m happypose.toolbox.utils.download --cosypose_models \
-          detector-bop-ycbv-pbr--970850 \
-          coarse-bop-ycbv-pbr--724183 \
-          refiner-bop-ycbv-pbr--604090
-
+python -m happypose.toolbox.utils.download --cosypose_models detector-bop-ycbv-pbr--970850
 # tless
-python -m happypose.toolbox.utils.download --cosypose_models \
-          detector-bop-tless-pbr--873074 \
-          coarse-bop-tless-pbr--506801 \
-          refiner-bop-tless-pbr--233420
+python -m happypose.toolbox.utils.download --cosypose_models detector-bop-tless-pbr--873074
 ```
 
 # Dataset

docs/book/megapose/evaluate.md

@@ -2,28 +2,11 @@
 
 Please make sure you followed the steps relative to the evaluation in the main readme.
 
-## TEASER++
-
-You will first need to install [TEASER++](https://github.com/MIT-SPARK/TEASER-plusplus) if you want to use the depth for MegaPose.
-To do so, please run the following commands to install it :
-
-```
-# Go to HappyPose root directory
-apt install -y cmake libeigen3-dev libboost-all-dev
-conda activate happypose
-mamba install compilers -c conda-forge
-pip install open3d
-mkdir /build && cd /build && git clone https://github.com/MIT-SPARK/TEASER-plusplus.git
-cd TEASER-plusplus && mkdir build && cd build
-cmake -DTEASERPP_PYTHON_VERSION=3.9 .. && make teaserpp_python
-cd python && pip install .
-```
-
 ## Evaluating with Megapose detector
 Run a detector part of Megapose pipeline to detect bounding boxes in the image dataset at run-time.
 
 ```
-python -m happypose.pose_estimators.megapose.scripts.run_full_megapose_eval detector_run_id=bop_pbr coarse_run_id=coarse-rgb-906902141 refiner_run_id=refiner-rgb-653307694 ds_names=[ycbv.bop19,lmo.bop19,tless.bop19,tudl.bop19,icbin.bop19,hb.bop19,itodd.bop19] result_id=detector_1posehyp detection_coarse_types=[["detector","SO3_grid"]] inference.n_pose_hypotheses=1 skip_inference=false run_bop_eval=true
+python -m happypose.pose_estimators.megapose.scripts.run_full_megapose_eval detector_run_id=bop_pbr coarse_run_id=coarse-rgb-906902141 refiner_run_id=refiner-rgb-653307694 ds_names=[ycbv.bop19,lmo.bop19,tless.bop19,tudl.bop19,icbin.bop19,hb.bop19,itodd.bop19,hope.bop19] result_id=detector_1posehyp detection_coarse_types=[["detector","SO3_grid"]] inference.n_pose_hypotheses=1 skip_inference=false run_bop_eval=true
 ```
 
 ## Evaluating with external detections

environment.yml

@@ -11,6 +11,7 @@ dependencies:
   - pip
   - pytorch::pytorch==1.11.0
   - torchvision==0.12.0
+  - mkl==2024.0.0
   - geckodriver
   - firefox
   - opencv

happypose/pose_estimators/cosypose/cosypose/config.py

@@ -25,8 +25,9 @@
 SYNT_DS_DIR = LOCAL_DATA_DIR / "synt_datasets"
 BOP_DS_DIR = LOCAL_DATA_DIR / "bop_datasets"
 
-BOP_TOOLKIT_DIR = PROJECT_DIR / "deps" / "bop_toolkit_cosypose"
-BOP_CHALLENGE_TOOLKIT_DIR = PROJECT_DIR / "deps" / "bop_toolkit_challenge"
+# BOP scripts
+BOP_POSE_EVAL_SCRIPT_NAME = "eval_bop19_pose.py"
+BOP_DETECTION_EVAL_SCRIPT_NAME = "eval_bop22_coco.py"
 
 EXP_DIR = LOCAL_DATA_DIR / "experiments"
 RESULTS_DIR = LOCAL_DATA_DIR / "results"

happypose/pose_estimators/cosypose/cosypose/datasets/bop.py

@@ -1,23 +1,17 @@
 import json
 import pickle
-import sys
 from pathlib import Path
 
 import numpy as np
 import pandas as pd
 import torch
+from bop_toolkit_lib import inout
 from PIL import Image
 
-from happypose.pose_estimators.cosypose.cosypose.config import BOP_TOOLKIT_DIR, MEMORY
+from happypose.pose_estimators.cosypose.cosypose.config import MEMORY
 from happypose.pose_estimators.cosypose.cosypose.utils.logging import get_logger
 from happypose.toolbox.lib3d.transform import Transform
 
-sys.path.append(str(BOP_TOOLKIT_DIR))
-from bop_toolkit_lib import inout
-
-sys.path = sys.path[:-1]
-
-
 logger = get_logger(__name__)
 
 

happypose/pose_estimators/cosypose/cosypose/evaluation/evaluation.py

@@ -42,12 +42,8 @@
     ModelNetErrorMeter,
 )
 from happypose.pose_estimators.megapose.evaluation.runner_utils import format_results
-from happypose.pose_estimators.megapose.inference.icp_refiner import ICPRefiner
 
 # Pose estimator
-from happypose.pose_estimators.megapose.inference.teaserpp_refiner import (
-    TeaserppRefiner,
-)
 from happypose.toolbox.datasets.datasets_cfg import make_object_dataset
 from happypose.toolbox.lib3d.rigid_mesh_database import MeshDataBase
 from happypose.toolbox.renderer.panda3d_batch_renderer import Panda3dBatchRenderer
@@ -189,22 +185,7 @@ def run_eval(
     # See https://stackoverflow.com/a/53287330
     assert cfg.coarse_run_id is not None
     assert cfg.refiner_run_id is not None
-    # TODO (emaitre): This fuction seems to take the wrong parameters. Trying to fix
-    # this.
-    """
-    (
-        coarse_model,
-        refiner_model,
-        mesh_db,
-    ) = happypose.toolbox.inference.utils.load_pose_models(
-        coarse_run_id=cfg.coarse_run_id,
-        refiner_run_id=cfg.refiner_run_id,
-        n_workers=cfg.n_rendering_workers,
-        obj_ds_name=obj_ds_name,
-        urdf_ds_name=urdf_ds_name,
-        force_panda3d_renderer=True,
-    )
-    """
+
     object_ds = make_object_dataset(obj_ds_name)
 
     coarse_model, refiner_model, mesh_db = load_pose_models(
@@ -217,8 +198,16 @@ def run_eval(
 
     if cfg.inference.run_depth_refiner:
         if cfg.inference.depth_refiner == "icp":
+            from happypose.pose_estimators.megapose.inference.icp_refiner import (
+                ICPRefiner,
+            )
+
             ICPRefiner(mesh_db, renderer)
         elif cfg.inference.depth_refiner == "teaserpp":
+            from happypose.pose_estimators.megapose.inference.teaserpp_refiner import (
+                TeaserppRefiner,
+            )
+
             TeaserppRefiner(mesh_db, renderer)
         else:
             pass

happypose/pose_estimators/cosypose/cosypose/evaluation/prediction_runner.py

@@ -173,46 +173,43 @@ def get_predictions(
         """
         predictions_list = defaultdict(list)
         for n, data in enumerate(tqdm(self.dataloader)):
-            if n < 3:
-                # data is a dict
-                rgb = data["rgb"]
-                depth = None
-                K = data["cameras"].K
-                gt_detections = data["gt_detections"].cuda()
-
-                initial_data = None
-                if data["initial_data"]:
-                    initial_data = data["initial_data"].cuda()
-
-                obs_tensor = ObservationTensor.from_torch_batched(rgb, depth, K)
-                obs_tensor = obs_tensor.cuda()
-
-                # GPU warmup for timing
-                if n == 0:
-                    with torch.no_grad():
-                        self.run_inference_pipeline(
-                            pose_estimator,
-                            obs_tensor,
-                            gt_detections,
-                            initial_estimates=initial_data,
-                        )
-
-                cuda_timer = CudaTimer()
-                cuda_timer.start()
+            # data is a dict
+            rgb = data["rgb"]
+            depth = None
+            K = data["cameras"].K
+            gt_detections = data["gt_detections"].cuda()
+
+            initial_data = None
+            if data["initial_data"]:
+                initial_data = data["initial_data"].cuda()
+
+            obs_tensor = ObservationTensor.from_torch_batched(rgb, depth, K)
+            obs_tensor = obs_tensor.cuda()
+
+            # GPU warmup for timing
+            if n == 0:
                 with torch.no_grad():
-                    all_preds = self.run_inference_pipeline(
+                    self.run_inference_pipeline(
                         pose_estimator,
                         obs_tensor,
                         gt_detections,
                         initial_estimates=initial_data,
                     )
-                cuda_timer.end()
-                cuda_timer.elapsed()
 
-                for k, v in all_preds.items():
-                    predictions_list[k].append(v)
-            else:
-                break
+            cuda_timer = CudaTimer()
+            cuda_timer.start()
+            with torch.no_grad():
+                all_preds = self.run_inference_pipeline(
+                    pose_estimator,
+                    obs_tensor,
+                    gt_detections,
+                    initial_estimates=initial_data,
+                )
+            cuda_timer.end()
+            cuda_timer.elapsed()
+
+            for k, v in all_preds.items():
+                predictions_list[k].append(v)
 
         # Concatenate the lists of PandasTensorCollections
         predictions = {}

happypose/pose_estimators/cosypose/cosypose/scripts/run_bop20_eval.py

@@ -1,23 +1,15 @@
 import argparse
-import os
 import shutil
 import subprocess
-import sys
-from pathlib import Path
 
 import torch
+from bop_toolkit_lib import inout
 from tqdm import tqdm
 
-from happypose.pose_estimators.cosypose.cosypose.config import PROJECT_DIR, RESULTS_DIR
-
-TOOLKIT_DIR = Path(PROJECT_DIR / "deps" / "bop_toolkit_challenge")
-EVAL_SCRIPT_PATH = TOOLKIT_DIR / "scripts/eval_bop19.py"
-DUMMY_EVAL_SCRIPT_PATH = TOOLKIT_DIR / "scripts/eval_bop19_dummy.py"
-
-sys.path.append(TOOLKIT_DIR.as_posix())
-from bop_toolkit_lib import inout  # noqa
-
-# from bop_toolkit_lib.config import results_path as BOP_RESULTS_PATH
+from happypose.pose_estimators.cosypose.cosypose.config import (
+    BOP_POSE_EVAL_SCRIPT_NAME,
+    RESULTS_DIR,
+)
 
 
 def main():
@@ -47,7 +39,7 @@ def run_evaluation(args):
         )
 
     if not args.convert_only:
-        run_bop_evaluation(csv_path, dummy=args.dummy)
+        run_bop_evaluation(csv_path)
     return csv_path
 
 
@@ -82,25 +74,15 @@ def convert_results(results_path, out_csv_path, method):
     return out_csv_path
 
 
-def run_bop_evaluation(filename, dummy=False):
-    myenv = os.environ.copy()
-    myenv["PYTHONPATH"] = TOOLKIT_DIR.as_posix()
-    myenv["COSYPOSE_DIR"] = PROJECT_DIR.as_posix()
-    if dummy:
-        script_path = DUMMY_EVAL_SCRIPT_PATH
-    else:
-        script_path = EVAL_SCRIPT_PATH
+def run_bop_evaluation(filename):
     subprocess.call(
         [
-            "python",
-            script_path.as_posix(),
+            BOP_POSE_EVAL_SCRIPT_NAME,
             "--renderer_type",
             "python",
             "--result_filenames",
             filename,
         ],
-        env=myenv,
-        cwd=TOOLKIT_DIR.as_posix(),
     )