feat: add other video examples

app.py

@@ -1,10 +1,11 @@
 """
-Gradio app to showcase the pyronear model for early forest fire detection.
+Gradio app to showcase the pyronear model for salmon vision.
 """
 
 from collections import Counter
 from pathlib import Path
 from typing import Any, Tuple
+import torch
 
 import gradio as gr
 import numpy as np
@@ -18,6 +19,18 @@ def bgr_to_rgb(a: np.ndarray) -> np.ndarray:
     """
     return a[:, :, ::-1]
 
+def has_values(maybe_tensor: torch.Tensor | None) -> bool:
+    """
+    Check whether the `maybe_tensor` contains items.
+    """
+    if maybe_tensor is None:
+        return False
+    elif isinstance(maybe_tensor, torch.Tensor):
+        if len(maybe_tensor) == 0:
+            return False
+        else:
+            return True
+
 
 def analyze_predictions(yolo_predictions) -> dict[str, Any]:
     """
@@ -43,14 +56,14 @@ def analyze_predictions(yolo_predictions) -> dict[str, Any]:
         names = yolo_predictions[0].names
         ids = set()
         for prediction in yolo_predictions:
-            if prediction.boxes.id:
+            if has_values(prediction.boxes.id):
                 for id in prediction.boxes.id.numpy().astype("int"):
                     ids.add(id.item())
         detected_species = {}
         for id in ids:
             counter = Counter()
             for prediction in yolo_predictions:
-                if prediction.boxes.id:
+                if has_values(prediction.boxes.id):
                     for idd, klass in zip(
                         prediction.boxes.id.numpy().astype("int"),
                         prediction.boxes.cls.numpy().astype("int"),
@@ -88,7 +101,7 @@ def prediction_to_str(yolo_predictions) -> str:
         return f"Detected {len(ids)} salmons in the video clip with ids {ids}:\n{summary_str}"
 
 
-def predict(model: YOLO, video_filepath: Path) -> Tuple[Path, str]:
+def interface_fn(model: YOLO, video_filepath: Path) -> Tuple[Path, str]:
     """
     Main interface function that runs the model on the provided pil_image and
     returns the exepected tuple to populate the gradio interface.
@@ -161,7 +174,7 @@ with gr.Blocks() as demo:
     )
     output_raw = gr.Text(label="raw prediction")
 
-    fn = lambda video_filepath: predict(
+    fn = lambda video_filepath: interface_fn(
         model=model, video_filepath=Path(video_filepath)
     )
     gr.Interface(

data/videos/{video1-clip.mp4 → video1-clip-fps-20.mp4}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3fb22eeeb3be60b9cb2a65d54a2d6c5379e9e65c47d7f2ce88eab0986f069167
-size 2966504
+oid sha256:be2171afaba6acd9acd68138aa6d02334b5f2e711103383dcc984d05dd87232b
+size 1753568

data/videos/video2-clip-fps-20.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bcc2bf599356da9d9db57f05d17f72f2366e0c7f63c0bc562fe5a34a7ac757c6
+size 1078207

data/videos/video3-clip-fps-20.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:808290484ba8088aae55b4af4faf48f45b6efe9524b2d0b3c85d26d6a95ad59b
+size 3196588