summarize stats

examples/benchmarks/compression/mcmc.sh

@@ -49,7 +49,7 @@ done
 if command -v zip &> /dev/null
 then
     echo "Zipping results"
-    python benchmarks/compression/summarize_stats.py --results_dir $RESULT_DIR
+    python benchmarks/summarize_stats.py --results_dir $RESULT_DIR --scenes $SCENE_LIST --stage compress
 else
     echo "zip command not found, skipping zipping"
 fi

examples/benchmarks/compression/mcmc_tt.sh

@@ -42,7 +42,7 @@ done
 if command -v zip &> /dev/null
 then
     echo "Zipping results"
-    python benchmarks/compression/summarize_stats.py --results_dir $RESULT_DIR --scenes $SCENE_LIST
+    python benchmarks/summarize_stats.py --results_dir $RESULT_DIR --scenes $SCENE_LIST --stage compress
 else
     echo "zip command not found, skipping zipping"
 fi

examples/benchmarks/normal/2dgs_dtu.sh

@@ -0,0 +1,17 @@
+SCENE_DIR="data/DTU"
+SCENE_LIST="scan24 scan37 scan40 scan55 scan63 scan65 scan69 scan83 scan97 scan105 scan106 scan110 scan114 scan118 scan122"
+
+RESULT_DIR="results/benchmark_dtu_2dgs"
+
+for SCENE in $SCENE_LIST;
+do
+    echo "Running $SCENE"
+
+    # train and eval
+    CUDA_VISIBLE_DEVICES=0 python simple_trainer_2dgs.py --disable_viewer --data_factor 1 \
+        --data_dir $SCENE_DIR/$SCENE/ \
+        --result_dir $RESULT_DIR/$SCENE/
+done
+
+echo "Summarizing results"
+python benchmarks/summarize_stats.py --results_dir $RESULT_DIR --scenes $SCENE_LIST --stage val

examples/benchmarks/normal/mcmc_dtu.sh

@@ -0,0 +1,28 @@
+SCENE_DIR="data/DTU"
+SCENE_LIST="scan24 scan37 scan40 scan55 scan63 scan65 scan69 scan83 scan97 scan105 scan106 scan110 scan114 scan118 scan122"
+RENDER_TRAJ_PATH="ellipse"
+
+RESULT_DIR="results/benchmark_dtu_mcmc_0.25M_normal"
+CAP_MAX=250000
+
+# RESULT_DIR="results/benchmark_dtu_mcmc_0.5M_normal"
+# CAP_MAX=500000
+
+# RESULT_DIR="results/benchmark_dtu_mcmc_1M_normal"
+# CAP_MAX=1000000
+
+for SCENE in $SCENE_LIST;
+do
+    echo "Running $SCENE"
+
+    # train and eval
+    CUDA_VISIBLE_DEVICES=0 python simple_trainer.py mcmc --disable_viewer --data_factor 1 \
+        --strategy.cap-max $CAP_MAX \
+        --normal_consistency_loss \
+        --render_traj_path $RENDER_TRAJ_PATH \
+        --data_dir $SCENE_DIR/$SCENE/ \
+        --result_dir $RESULT_DIR/$SCENE/
+done
+
+echo "Summarizing results"
+python benchmarks/summarize_stats.py --results_dir $RESULT_DIR --scenes $SCENE_LIST --stage val

examples/benchmarks/normal/mcmc_normal.sh

@@ -1,5 +1,5 @@
 SCENE_DIR="data/360_v2"
-SCENE_LIST="garden bicycle stump bonsai counter kitchen room" # treehill flowers
+SCENE_LIST="garden bicycle stump bonsai counter kitchen room treehill flowers"
 
 RESULT_DIR="results/benchmark_normal"
 RENDER_TRAJ_PATH="ellipse"
@@ -20,3 +20,6 @@ do
         --data_dir $SCENE_DIR/$SCENE/ \
         --result_dir $RESULT_DIR/$SCENE/
 done
+
+echo "Summarizing results"
+python benchmarks/summarize_stats.py --results_dir $RESULT_DIR --scenes $SCENE_LIST --stage val

examples/benchmarks/compression/summarize_stats.py → examples/benchmarks/summarize_stats.py

@@ -8,11 +8,8 @@
 import tyro
 
 
-def main(results_dir: str, scenes: List[str]):
-    print("scenes:", scenes)
-    stage = "compress"
-
-    summary = defaultdict(list)
+def main(results_dir: str, scenes: List[str], stage: str = "val"):
+    stats_all = defaultdict(list)
     for scene in scenes:
         scene_dir = os.path.join(results_dir, scene)
 
@@ -25,15 +22,20 @@ def main(results_dir: str, scenes: List[str]):
                 f"stat -c%s {zip_path}", shell=True, capture_output=True
             )
             size = int(out.stdout)
-            summary["size"].append(size)
+            stats_all["size"].append(size)
 
         with open(os.path.join(scene_dir, f"stats/{stage}_step29999.json"), "r") as f:
             stats = json.load(f)
             for k, v in stats.items():
-                summary[k].append(v)
+                stats_all[k].append(v)
+
+    summary = {"scenes": scenes}
+    for k, v in stats_all.items():
+        summary[k] = np.mean(v)
+    print(summary)
 
-    for k, v in summary.items():
-        print(k, np.mean(v))
+    with open(os.path.join(results_dir, f"{stage}_summary.json"), "w") as f:
+        json.dump(summary, f, indent=2)
 
 
 if __name__ == "__main__":