ivangrov / yolov3-googlecolab

A walk through the code behind setting up YOLOv3 with darknet and training it and processing video on Google Colaboratory

Hey there, I'm Ivan Goncharov 👋

About me

• 🔭 I’m currently helping build the best tools for machine learning practitioners at Weights & Biases.
• 🎥 As well as creating educational videos about ML on my YouTube channel.
• 🌱 I’m currently learning more about domains of computer vision, NLP and MLOps.
• 👯 I’m happy to collaborate on any exciting ML projects.
• 📫 How to reach me: You can reach me on my Twitter and my LinkedIn. You can email me at [email protected] / [email protected].
• ⚡ Fun fact: I'm really into music and enjoy spending my time jamming on the guitar.

Github Stats

yolov3-googlecolab's Issues

Unable to open file (file signature not found )

`---------------------------------------------------------------------------
OSError Traceback (most recent call last)
in ()
----> 1 yolo_model = load_model("drive/My Drive/car_yolo/car_detection/model_data/yolo.h5")

3 frames
/usr/local/lib/python3.6/dist-packages/h5py/_hl/files.py in make_fid(name, mode, userblock_size, fapl, fcpl, swmr)
171 if swmr and swmr_support:
172 flags |= h5f.ACC_SWMR_READ
--> 173 fid = h5f.open(name, flags, fapl=fapl)
174 elif mode == 'r+':
175 fid = h5f.open(name, h5f.ACC_RDWR, fapl=fapl)

h5py/_objects.pyx in h5py._objects.with_phil.wrapper()

h5py/_objects.pyx in h5py._objects.with_phil.wrapper()

h5py/h5f.pyx in h5py.h5f.open()

OSError: Unable to open file (file signature not found)`

Can't detect Custom Objects

Hello!
I am trying to detect my custom objects but my YOLO network traning process takes a lot of time. I am doing it with Google Colab and it loses its connection before the convergence of the loss function (after 6000 iterations) so it can 't detect any object...

Do you know how can I solve that problem?

Also I have a doubt that maybe that's the reason of my problem: When I upload the training images with their corresponding bounding box I don't use entire images, I just upload images of the object I want to detect (with their bounding box):

i.e.

instead of:

Could it be the reason of my problem?

Thank you so much

Got ValueError: could not convert string to float: '/home/jovyan/work_sro/week3/Car_detection_for_Autonomous_Driving/model_data/yolo_anchors.txt'

yolo_model = load_model("drive/My Drive/car_yolo/car_detection/model_data/yolo.h5")

Transfer learning

Hello! Is it possible to fit the network on my own class of objects or is it necessary to train from scratch here?

Couldnot convert string to float in the anchor file @Object Detection

class_names = read_classes("coco_classes.txt") anchors = read_anchors("yolo_anchors.txt") image_shape = (720., 1280.)

Got error like this:

ValueError Traceback (most recent call last)
in ()
1 class_names = read_classes("coco_classes.txt")
----> 2 anchors = read_anchors("yolo_anchors.txt")
3 image_shape = (720., 1280.)
`/content/yolo_utils.py in (.0)

 17     with open(anchors_path) as f:
 18         anchors = f.readlines()

---> 19 anchors = np.loadtxt(anchors_path, delimiter=",")
20 anchors = [float(x) for x in anchors.split(",")]
21 anchors = np.array(anchors).reshape(-1, 2)

ValueError: could not convert string to float: '/home/jovyan/work_sro/week3/Car_`

Can't train YOLO in GoogleColab

Hi Ivan!
I'm new in Google Colab and I'm having some problems to train my YOLO network with my custom data.

I am executing:
!./darknet detector train data/obj.data yolov3-tiny-obj.cfg yolov3-tiny.conv.15 -dont_show
inside x64 folder and I figure the following error:

Please, can you help me?

Thank you so much and congrats for all your work :)

Software installation on Google Colab

Heey, thank you for your detailed help. I want to ask you, if I install these packages on Google Drive you know for training a dataset, do they go away after 15 hours as well ?

bbox coordinates and yaml

Hi Ivan,
In the Bbox label tool you use, the bbox text file contains for e.g.
0 0.48 0.37 0.009 0.027

0 is the class, what are the other values? How does it relate to a yaml (xmin, ymin, xmax, ymax) like like:

      <xmin>853.1795089352</xmin>
      <ymin>289.0164594851</ymin>
      <xmax>860.6634561323</xmax>
      <ymax>305.5028603386</ymax>

Thanks