Autonomous Driving¶
Run in Google ColabObjective: Use the "You Only Look Once" (YOLO) algorithm to build a car detection system.
You Only Look Once (YOLO) is a popular real-time object detection system that achieves high accuracy by requiring only one forward propagation through the network to make predictions. It recognizes objects along with their corresponding bounding boxes.
Download the "Yet Another Darknet 2 Keras" (YAD2K) repository¶
Yet Another Darknet 2 Keras (YAD2K) is a repository that converts the Darknet YOLOv2 model to a Keras model. The YAD2K code was modified to work with Keras 3.
In [1]:
%%bash
if [ -e "/tmp/YAD2K.zip" ]; then
echo "YAD2K.zip already exists!"
else
gdown 1MA4g7d-pJEChYey2PotscZXq9EJr0K2H -O /tmp/
fi
unzip -qn /tmp/YAD2K.zip -d /tmp
YAD2K.zip already exists!
Import libraries¶
In [2]:
import sys
sys.path.insert(0, '/tmp/YAD2K')
from yad2k.utils.utils import preprocess_image, read_classes, read_anchors, draw_boxes
from yad2k.models.keras_yolo import space_to_depth_x2, space_to_depth_x2_output_shape, yolo_head, yolo_eval
import os
from keras import models
import matplotlib.pyplot as plt
from PIL import Image
import matplotlib.animation as animation
from IPython.display import HTML
2025-01-04 12:24:40.493598: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:477] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered WARNING: All log messages before absl::InitializeLog() is called are written to STDERR E0000 00:00:1736015080.519247 102025 cuda_dnn.cc:8310] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1736015080.528497 102025 cuda_blas.cc:1418] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered 2025-01-04 12:24:40.553545: I tensorflow/core/platform/cpu_feature_guard.cc:210] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations. To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
Visualize the dataset¶
In [3]:
path = "/tmp/YAD2K/images"
files = [os.path.join(path, file) for file in sorted(os.listdir(path)) if file.endswith(".jpg") and file.startswith("0")]
frames = []
fig, ax = plt.subplots()
for file in files:
image = Image.open(file)
plt.tight_layout()
ax.set_title("Frames")
ax.axis("off")
frame = ax.imshow(image, animated=True)
frames.append([frame])
anim = animation.ArtistAnimation(fig, frames, blit=True, repeat_delay=1000)
plt.close(fig)
HTML(anim.to_html5_video())
Out[3]:
Download the Darknet YOLOv2 model and convert it to a Keras model¶
In [4]:
%%bash
wget -nc --progress=bar:force:noscroll https://pjreddie.com/media/files/yolov2.weights -P /tmp/YAD2K/model_data
wget -nc --progress=bar:force:noscroll https://raw.githubusercontent.com/pjreddie/darknet/master/cfg/yolov2.cfg -P /tmp/YAD2K/model_data
if [ -e "/tmp/YAD2K/model_data/yolov2.keras" ]; then
echo "yolov2.keras already exists!"
else
/tmp/YAD2K/yad2k.py /tmp/YAD2K/model_data/yolov2.cfg /tmp/YAD2K/model_data/yolov2.weights /tmp/YAD2K/model_data/yolov2.keras
fi
File '/tmp/YAD2K/model_data/yolov2.weights' already there; not retrieving. File '/tmp/YAD2K/model_data/yolov2.cfg' already there; not retrieving.
yolov2.keras already exists!
Load the Keras model, classes, and anchor boxes¶
In [ ]:
classes = read_classes("/tmp/YAD2K/model_data/coco_classes.txt")
anchors = read_anchors("/tmp/YAD2K/model_data/yolov2_anchors.txt")
model = models.load_model("/tmp/YAD2K/model_data/yolov2.keras",
custom_objects={'space_to_depth_x2': space_to_depth_x2, 'space_to_depth_x2_output_shape': space_to_depth_x2_output_shape})
model.summary()
I0000 00:00:1736015107.116582 102025 gpu_device.cc:2022] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 1624 MB memory: -> device: 0, name: NVIDIA GeForce GTX 1650, pci bus id: 0000:01:00.0, compute capability: 7.5
Model: "functional"
┏━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┓ ┃ Layer (type) ┃ Output Shape ┃ Param # ┃ Connected to ┃ ┡━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━┩ │ input_layer │ (None, 608, 608, │ 0 │ - │ │ (InputLayer) │ 3) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d (Conv2D) │ (None, 608, 608, │ 864 │ input_layer[0][0] │ │ │ 32) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalization │ (None, 608, 608, │ 128 │ conv2d[0][0] │ │ (BatchNormalizatio… │ 32) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu │ (None, 608, 608, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 32) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ max_pooling2d │ (None, 304, 304, │ 0 │ leaky_re_lu[0][0] │ │ (MaxPooling2D) │ 32) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_1 (Conv2D) │ (None, 304, 304, │ 18,432 │ max_pooling2d[0]… │ │ │ 64) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 304, 304, │ 256 │ conv2d_1[0][0] │ │ (BatchNormalizatio… │ 64) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_1 │ (None, 304, 304, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 64) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ max_pooling2d_1 │ (None, 152, 152, │ 0 │ leaky_re_lu_1[0]… │ │ (MaxPooling2D) │ 64) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_2 (Conv2D) │ (None, 152, 152, │ 73,728 │ max_pooling2d_1[… │ │ │ 128) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 152, 152, │ 512 │ conv2d_2[0][0] │ │ (BatchNormalizatio… │ 128) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_2 │ (None, 152, 152, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 128) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_3 (Conv2D) │ (None, 152, 152, │ 8,192 │ leaky_re_lu_2[0]… │ │ │ 64) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 152, 152, │ 256 │ conv2d_3[0][0] │ │ (BatchNormalizatio… │ 64) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_3 │ (None, 152, 152, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 64) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_4 (Conv2D) │ (None, 152, 152, │ 73,728 │ leaky_re_lu_3[0]… │ │ │ 128) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 152, 152, │ 512 │ conv2d_4[0][0] │ │ (BatchNormalizatio… │ 128) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_4 │ (None, 152, 152, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 128) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ max_pooling2d_2 │ (None, 76, 76, │ 0 │ leaky_re_lu_4[0]… │ │ (MaxPooling2D) │ 128) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_5 (Conv2D) │ (None, 76, 76, │ 294,912 │ max_pooling2d_2[… │ │ │ 256) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 76, 76, │ 1,024 │ conv2d_5[0][0] │ │ (BatchNormalizatio… │ 256) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_5 │ (None, 76, 76, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 256) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_6 (Conv2D) │ (None, 76, 76, │ 32,768 │ leaky_re_lu_5[0]… │ │ │ 128) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 76, 76, │ 512 │ conv2d_6[0][0] │ │ (BatchNormalizatio… │ 128) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_6 │ (None, 76, 76, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 128) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_7 (Conv2D) │ (None, 76, 76, │ 294,912 │ leaky_re_lu_6[0]… │ │ │ 256) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 76, 76, │ 1,024 │ conv2d_7[0][0] │ │ (BatchNormalizatio… │ 256) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_7 │ (None, 76, 76, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 256) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ max_pooling2d_3 │ (None, 38, 38, │ 0 │ leaky_re_lu_7[0]… │ │ (MaxPooling2D) │ 256) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_8 (Conv2D) │ (None, 38, 38, │ 1,179,648 │ max_pooling2d_3[… │ │ │ 512) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 38, 38, │ 2,048 │ conv2d_8[0][0] │ │ (BatchNormalizatio… │ 512) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_8 │ (None, 38, 38, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 512) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_9 (Conv2D) │ (None, 38, 38, │ 131,072 │ leaky_re_lu_8[0]… │ │ │ 256) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 38, 38, │ 1,024 │ conv2d_9[0][0] │ │ (BatchNormalizatio… │ 256) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_9 │ (None, 38, 38, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 256) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_10 (Conv2D) │ (None, 38, 38, │ 1,179,648 │ leaky_re_lu_9[0]… │ │ │ 512) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 38, 38, │ 2,048 │ conv2d_10[0][0] │ │ (BatchNormalizatio… │ 512) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_10 │ (None, 38, 38, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 512) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_11 (Conv2D) │ (None, 38, 38, │ 131,072 │ leaky_re_lu_10[0… │ │ │ 256) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 38, 38, │ 1,024 │ conv2d_11[0][0] │ │ (BatchNormalizatio… │ 256) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_11 │ (None, 38, 38, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 256) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_12 (Conv2D) │ (None, 38, 38, │ 1,179,648 │ leaky_re_lu_11[0… │ │ │ 512) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 38, 38, │ 2,048 │ conv2d_12[0][0] │ │ (BatchNormalizatio… │ 512) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_12 │ (None, 38, 38, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 512) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ max_pooling2d_4 │ (None, 19, 19, │ 0 │ leaky_re_lu_12[0… │ │ (MaxPooling2D) │ 512) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_13 (Conv2D) │ (None, 19, 19, │ 4,718,592 │ max_pooling2d_4[… │ │ │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 19, 19, │ 4,096 │ conv2d_13[0][0] │ │ (BatchNormalizatio… │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_13 │ (None, 19, 19, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_14 (Conv2D) │ (None, 19, 19, │ 524,288 │ leaky_re_lu_13[0… │ │ │ 512) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 19, 19, │ 2,048 │ conv2d_14[0][0] │ │ (BatchNormalizatio… │ 512) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_14 │ (None, 19, 19, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 512) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_15 (Conv2D) │ (None, 19, 19, │ 4,718,592 │ leaky_re_lu_14[0… │ │ │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 19, 19, │ 4,096 │ conv2d_15[0][0] │ │ (BatchNormalizatio… │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_15 │ (None, 19, 19, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_16 (Conv2D) │ (None, 19, 19, │ 524,288 │ leaky_re_lu_15[0… │ │ │ 512) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 19, 19, │ 2,048 │ conv2d_16[0][0] │ │ (BatchNormalizatio… │ 512) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_16 │ (None, 19, 19, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 512) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_17 (Conv2D) │ (None, 19, 19, │ 4,718,592 │ leaky_re_lu_16[0… │ │ │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 19, 19, │ 4,096 │ conv2d_17[0][0] │ │ (BatchNormalizatio… │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_17 │ (None, 19, 19, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_18 (Conv2D) │ (None, 19, 19, │ 9,437,184 │ leaky_re_lu_17[0… │ │ │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 19, 19, │ 4,096 │ conv2d_18[0][0] │ │ (BatchNormalizatio… │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_20 (Conv2D) │ (None, 38, 38, │ 32,768 │ leaky_re_lu_12[0… │ │ │ 64) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_18 │ (None, 19, 19, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 38, 38, │ 256 │ conv2d_20[0][0] │ │ (BatchNormalizatio… │ 64) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_19 (Conv2D) │ (None, 19, 19, │ 9,437,184 │ leaky_re_lu_18[0… │ │ │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_20 │ (None, 38, 38, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 64) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 19, 19, │ 4,096 │ conv2d_19[0][0] │ │ (BatchNormalizatio… │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ space_to_depth_x2 │ (None, 19, 19, │ 0 │ leaky_re_lu_20[0… │ │ (Lambda) │ 256) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_19 │ (None, 19, 19, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ concatenate │ (None, 19, 19, │ 0 │ space_to_depth_x… │ │ (Concatenate) │ 1280) │ │ leaky_re_lu_19[0… │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_21 (Conv2D) │ (None, 19, 19, │ 11,796,480 │ concatenate[0][0] │ │ │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ batch_normalizatio… │ (None, 19, 19, │ 4,096 │ conv2d_21[0][0] │ │ (BatchNormalizatio… │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ leaky_re_lu_21 │ (None, 19, 19, │ 0 │ batch_normalizat… │ │ (LeakyReLU) │ 1024) │ │ │ ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ │ conv2d_22 (Conv2D) │ (None, 19, 19, │ 435,625 │ leaky_re_lu_21[0… │ │ │ 425) │ │ │ └─────────────────────┴───────────────────┴────────────┴───────────────────┘
Total params: 50,983,561 (194.49 MB)
Trainable params: 50,962,889 (194.41 MB)
Non-trainable params: 20,672 (80.75 KB)
Evaluate the YOLOv2 model¶
In [7]:
file = "/tmp/YAD2K/images/test.jpg"
image, image_data = preprocess_image(file, model_image_size=(608, 608))
model_outputs = model(image_data)
outputs = yolo_head(model_outputs, anchors, len(classes))
out_boxes, out_scores, out_classes = yolo_eval(outputs, [image.size[1], image.size[0]], score_threshold=0.3, iou_threshold=0.5)
output_image = draw_boxes(image.copy(), out_boxes, out_classes, classes, out_scores)
fig, axs = plt.subplots(1, 2, figsize=(15, 5))
axs[0].imshow(image)
axs[0].axis("off")
axs[0].set_title("Input image")
axs[1].imshow(output_image)
axs[1].axis("off")
axs[1].set_title("Output image")
plt.show()
I0000 00:00:1736015109.109839 102025 cuda_dnn.cc:529] Loaded cuDNN version 90300 2025-01-04 12:25:09.617921: W external/local_xla/xla/tsl/framework/bfc_allocator.cc:306] Allocator (GPU_0_bfc) ran out of memory trying to allocate 4.08GiB with freed_by_count=0. The caller indicates that this is not a failure, but this may mean that there could be performance gains if more memory were available. 2025-01-04 12:25:09.692711: W external/local_xla/xla/tsl/framework/bfc_allocator.cc:306] Allocator (GPU_0_bfc) ran out of memory trying to allocate 4.09GiB with freed_by_count=0. The caller indicates that this is not a failure, but this may mean that there could be performance gains if more memory were available. 2025-01-04 12:25:09.795716: W external/local_xla/xla/tsl/framework/bfc_allocator.cc:306] Allocator (GPU_0_bfc) ran out of memory trying to allocate 4.15GiB with freed_by_count=0. The caller indicates that this is not a failure, but this may mean that there could be performance gains if more memory were available. 2025-01-04 12:25:09.872998: W external/local_xla/xla/tsl/framework/bfc_allocator.cc:306] Allocator (GPU_0_bfc) ran out of memory trying to allocate 2.15GiB with freed_by_count=0. The caller indicates that this is not a failure, but this may mean that there could be performance gains if more memory were available. 2025-01-04 12:25:09.873065: W external/local_xla/xla/tsl/framework/bfc_allocator.cc:306] Allocator (GPU_0_bfc) ran out of memory trying to allocate 4.29GiB with freed_by_count=0. The caller indicates that this is not a failure, but this may mean that there could be performance gains if more memory were available. 2025-01-04 12:25:09.988016: W external/local_xla/xla/tsl/framework/bfc_allocator.cc:306] Allocator (GPU_0_bfc) ran out of memory trying to allocate 4.27GiB with freed_by_count=0. The caller indicates that this is not a failure, but this may mean that there could be performance gains if more memory were available. 2025-01-04 12:25:10.143910: W external/local_xla/xla/tsl/framework/bfc_allocator.cc:306] Allocator (GPU_0_bfc) ran out of memory trying to allocate 5.34GiB with freed_by_count=0. The caller indicates that this is not a failure, but this may mean that there could be performance gains if more memory were available.
In [ ]:
frames = []
fig, ax = plt.subplots()
for file in files:
image, image_data = preprocess_image(file, model_image_size=(608, 608))
model_outputs = model(image_data)
outputs = yolo_head(model_outputs, anchors, len(classes))
out_boxes, out_scores, out_classes = yolo_eval(outputs, [image.size[1], image.size[0]], score_threshold=0.3, iou_threshold=0.5)
output_image = draw_boxes(image.copy(), out_boxes, out_classes, classes, out_scores)
plt.tight_layout()
ax.set_title("Output frames")
ax.axis("off")
frame = ax.imshow(output_image, animated=True)
frames.append([frame])
anim = animation.ArtistAnimation(fig, frames, blit=True, repeat_delay=1000)
plt.close(fig)
HTML(anim.to_html5_video())
Out[ ]: