Add complete self-contained setup script with checkpoint downloads

setup_vine_complete.sh

@@ -0,0 +1,335 @@
+#!/bin/bash
+set -e  # Exit on any error
+
+echo "============================================================================"
+echo "VINE Model - Complete Setup Script"
+echo "This script sets up everything needed to use the VINE model from HuggingFace"
+echo "Model: https://huggingface.co/video-fm/vine"
+echo "============================================================================"
+
+# Get the directory where this script is located
+SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+cd "$SCRIPT_DIR"
+
+# ============================================================================
+# Step 1: Create Conda Environment
+# ============================================================================
+echo ""
+echo "=========================================="
+echo "Step 1: Creating conda environment 'vine_demo' with Python 3.10..."
+echo "=========================================="
+conda create -n vine_demo python=3.10 -y
+
+# Activate environment
+echo ""
+echo "Activating environment..."
+source $(conda info --base)/etc/profile.d/conda.sh
+conda activate vine_demo
+
+# ============================================================================
+# Step 2: Install PyTorch
+# ============================================================================
+echo ""
+echo "=========================================="
+echo "Step 2: Installing PyTorch 2.7.1 with CUDA 12.6 support..."
+echo "=========================================="
+pip install torch==2.7.1 torchvision==0.22.1 --index-url https://download.pytorch.org/whl/cu126
+
+# Upgrade pip
+pip install --upgrade pip
+
+# ============================================================================
+# Step 3: Install Core Dependencies
+# ============================================================================
+echo ""
+echo "=========================================="
+echo "Step 3: Installing core dependencies..."
+echo "=========================================="
+
+# Install transformers and HuggingFace tools
+pip install transformers>=4.40.0
+pip install huggingface-hub
+pip install safetensors
+pip install accelerate
+
+# Install video processing dependencies
+pip install opencv-python
+pip install pillow
+pip install matplotlib
+pip install seaborn
+pip install pandas
+pip install numpy
+pip install tqdm
+pip install scikit-learn
+
+# Install Gradio for demos (optional)
+pip install gradio
+
+echo "✓ Core dependencies installed"
+
+# ============================================================================
+# Step 4: Clone Required Repositories
+# ============================================================================
+echo ""
+echo "=========================================="
+echo "Step 4: Cloning required repositories..."
+echo "=========================================="
+
+mkdir -p src
+cd src
+
+# Clone SAM2
+if [ ! -d "video-sam2" ]; then
+    echo "Cloning video-sam2..."
+    git clone https://github.com/video-fm/video-sam2.git
+else
+    echo "video-sam2 already exists, skipping clone"
+fi
+
+# Clone GroundingDINO
+if [ ! -d "GroundingDINO" ]; then
+    echo "Cloning GroundingDINO..."
+    git clone https://github.com/video-fm/GroundingDINO.git
+else
+    echo "GroundingDINO already exists, skipping clone"
+fi
+
+# Clone LASER
+if [ ! -d "LASER" ]; then
+    echo "Cloning LASER..."
+    git clone https://github.com/kevinxuez/LASER.git
+else
+    echo "LASER already exists, skipping clone"
+fi
+
+# Clone vine_hf
+if [ ! -d "vine_hf" ]; then
+    echo "Cloning vine_hf..."
+    git clone https://github.com/kevinxuez/vine_hf.git
+else
+    echo "vine_hf already exists, skipping clone"
+fi
+
+echo "✓ Repositories cloned"
+
+# ============================================================================
+# Step 5: Install Packages in Editable Mode
+# ============================================================================
+echo ""
+echo "=========================================="
+echo "Step 5: Installing packages in editable mode..."
+echo "=========================================="
+
+echo "Installing video-sam2..."
+pip install --no-cache-dir -e ./video-sam2
+
+echo "Installing GroundingDINO..."
+pip install --no-cache-dir --use-pep517 -e ./GroundingDINO
+
+echo "Installing LASER..."
+pip install --no-cache-dir -e ./LASER
+
+echo "Installing vine_hf..."
+pip install --no-cache-dir -e ./vine_hf
+
+echo "✓ Packages installed"
+
+# ============================================================================
+# Step 6: Build GroundingDINO Extensions
+# ============================================================================
+echo ""
+echo "=========================================="
+echo "Step 6: Building GroundingDINO native extensions..."
+echo "=========================================="
+cd GroundingDINO
+python setup.py build_ext --force --inplace
+cd ..
+
+# Return to main directory
+cd "$SCRIPT_DIR"
+
+echo "✓ Extensions built"
+
+# ============================================================================
+# Step 7: Download Model Checkpoints
+# ============================================================================
+echo ""
+echo "=========================================="
+echo "Step 7: Downloading model checkpoints..."
+echo "=========================================="
+
+# Create checkpoints directory
+mkdir -p checkpoints
+cd checkpoints
+
+# Download SAM2 checkpoint (~149 MB)
+if [ ! -f "sam2_hiera_tiny.pt" ]; then
+    echo "Downloading SAM2 checkpoint (sam2_hiera_tiny.pt ~149 MB)..."
+    wget -q --show-progress https://dl.fbaipublicfiles.com/segment_anything_2/072824/sam2_hiera_tiny.pt
+    echo "✓ SAM2 checkpoint downloaded"
+else
+    echo "✓ SAM2 checkpoint already exists"
+fi
+
+# Download SAM2 config
+if [ ! -f "sam2_hiera_t.yaml" ]; then
+    echo "Downloading SAM2 config (sam2_hiera_t.yaml)..."
+    wget -q --show-progress https://raw.githubusercontent.com/facebookresearch/sam2/main/sam2/configs/sam2.1/sam2.1_hiera_t.yaml -O sam2_hiera_t.yaml
+    echo "✓ SAM2 config downloaded"
+else
+    echo "✓ SAM2 config already exists"
+fi
+
+# Download GroundingDINO checkpoint (~662 MB)
+if [ ! -f "groundingdino_swint_ogc.pth" ]; then
+    echo "Downloading GroundingDINO checkpoint (groundingdino_swint_ogc.pth ~662 MB)..."
+    wget -q --show-progress https://github.com/IDEA-Research/GroundingDINO/releases/download/v0.1.0-alpha/groundingdino_swint_ogc.pth
+    echo "✓ GroundingDINO checkpoint downloaded"
+else
+    echo "✓ GroundingDINO checkpoint already exists"
+fi
+
+# Download GroundingDINO config
+if [ ! -f "GroundingDINO_SwinT_OGC.py" ]; then
+    echo "Downloading GroundingDINO config (GroundingDINO_SwinT_OGC.py)..."
+    wget -q --show-progress https://raw.githubusercontent.com/IDEA-Research/GroundingDINO/main/groundingdino/config/GroundingDINO_SwinT_OGC.py
+    echo "✓ GroundingDINO config downloaded"
+else
+    echo "✓ GroundingDINO config already exists"
+fi
+
+# Return to main directory
+cd "$SCRIPT_DIR"
+
+echo ""
+echo "✓ All checkpoints downloaded to: $SCRIPT_DIR/checkpoints/"
+
+# ============================================================================
+# Step 8: Create Test Script
+# ============================================================================
+echo ""
+echo "=========================================="
+echo "Step 8: Creating test script..."
+echo "=========================================="
+
+cat > test_vine.py << 'TESTEOF'
+"""
+Test script for VINE model loaded from HuggingFace Hub
+"""
+import os
+import sys
+from pathlib import Path
+
+os.environ['OPENAI_API_KEY'] = "dummy-key"
+
+# Add src to path
+sys.path.insert(0, str(Path(__file__).parent / "src"))
+
+print("=" * 80)
+print("Testing VINE Model from video-fm/vine")
+print("=" * 80)
+
+# Load VINE from HuggingFace
+print("\n1. Loading VINE model from HuggingFace Hub...")
+from transformers import AutoModel
+model = AutoModel.from_pretrained('video-fm/vine', trust_remote_code=True)
+print("✓ Model loaded successfully")
+
+# Verify checkpoint files
+print("\n2. Verifying checkpoint files...")
+checkpoint_dir = Path(__file__).parent / "checkpoints"
+checkpoints = {
+    "SAM2 config": checkpoint_dir / "sam2_hiera_t.yaml",
+    "SAM2 checkpoint": checkpoint_dir / "sam2_hiera_tiny.pt",
+    "GroundingDINO config": checkpoint_dir / "GroundingDINO_SwinT_OGC.py",
+    "GroundingDINO checkpoint": checkpoint_dir / "groundingdino_swint_ogc.pth",
+}
+
+all_exist = True
+for name, path in checkpoints.items():
+    if path.exists():
+        size_mb = path.stat().st_size / (1024 * 1024)
+        print(f"✓ {name}: {path.name} ({size_mb:.1f} MB)")
+    else:
+        print(f"✗ {name}: NOT FOUND at {path}")
+        all_exist = False
+
+# Create pipeline
+print("\n3. Creating VINE pipeline...")
+from vine_hf import VinePipeline
+
+pipeline = VinePipeline(
+    model=model,
+    tokenizer=None,
+    sam_config_path=str(checkpoints["SAM2 config"]),
+    sam_checkpoint_path=str(checkpoints["SAM2 checkpoint"]),
+    gd_config_path=str(checkpoints["GroundingDINO config"]),
+    gd_checkpoint_path=str(checkpoints["GroundingDINO checkpoint"]),
+    device="cuda",
+    trust_remote_code=True
+)
+print("✓ Pipeline created successfully")
+
+print("\n" + "=" * 80)
+print("✅ VINE Setup Complete and Working!")
+print("=" * 80)
+print("\nYou can now use the model for video understanding:")
+print("""
+from transformers import AutoModel
+from vine_hf import VinePipeline
+
+model = AutoModel.from_pretrained('video-fm/vine', trust_remote_code=True)
+pipeline = VinePipeline(model=model, ...)
+results = pipeline('video.mp4', categorical_keywords=['person', 'dog'], ...)
+""")
+TESTEOF
+
+echo "✓ Test script created: test_vine.py"
+
+# ============================================================================
+# Step 9: Test the Installation
+# ============================================================================
+echo ""
+echo "=========================================="
+echo "Step 9: Testing installation..."
+echo "=========================================="
+
+echo "Checking PyTorch and CUDA..."
+python -c "import torch; print(f'PyTorch version: {torch.__version__}'); print(f'CUDA available: {torch.cuda.is_available()}'); print(f'CUDA version: {torch.version.cuda if torch.cuda.is_available() else \"N/A\"}')"
+
+echo ""
+echo "Running VINE model test..."
+python test_vine.py
+
+# ============================================================================
+# Final Summary
+# ============================================================================
+echo ""
+echo "============================================================================"
+echo "✅ VINE Setup Complete!"
+echo "============================================================================"
+echo ""
+echo "What was installed:"
+echo "  ✓ Conda environment: vine_demo"
+echo "  ✓ PyTorch 2.7.1 with CUDA 12.6"
+echo "  ✓ Required packages: laser, sam2, groundingdino, vine_hf"
+echo "  ✓ Model checkpoints downloaded to: checkpoints/"
+echo ""
+echo "Checkpoint files:"
+echo "  ✓ checkpoints/sam2_hiera_tiny.pt (~149 MB)"
+echo "  ✓ checkpoints/sam2_hiera_t.yaml"
+echo "  ✓ checkpoints/groundingdino_swint_ogc.pth (~662 MB)"
+echo "  ✓ checkpoints/GroundingDINO_SwinT_OGC.py"
+echo ""
+echo "To use VINE:"
+echo "  1. Activate environment: conda activate vine_demo"
+echo "  2. Run your script or test_vine.py"
+echo ""
+echo "Example usage:"
+echo "  python test_vine.py"
+echo ""
+echo "Model URL: https://huggingface.co/video-fm/vine"
+echo "Documentation: See README.md on HuggingFace Hub"
+echo ""
+echo "🎉 Happy video understanding with VINE!"
+echo "============================================================================"