Create read_rod_results.py

rodnet/datasets/loaders/read_rod_results.py

+import math
+
+from rodnet.core.object_class import get_class_id
+
+
+def load_rodnet_res(filename, config_dict):
+    n_class = config_dict['class_cfg']['n_class']
+    classes = config_dict['class_cfg']['classes']
+    model_configs = config_dict['model_cfg']
+    rng_grid = config_dict['mappings']['range_grid']
+    agl_grid = config_dict['mappings']['angle_grid']
+    test_configs = config_dict['test_cfg']
+
+    with open(filename, 'r') as df:
+        data = df.readlines()
+    if len(data) == 0:
+        return None, 0
+
+    n_frame = int(float(data[-1].rstrip().split()[0])) + 1
+    dts = {(i, j): [] for i in range(n_frame) for j in range(n_class)}
+
+    for id, line in enumerate(data):
+        if line is not None:
+            line = line.rstrip().split()
+            frameid, class_str, ridx, aidx, conf = line
+            frameid = int(frameid)
+            classid = get_class_id(class_str, classes)
+            ridx = int(ridx)
+            aidx = int(aidx)
+            conf = float(conf)
+            if conf > 1:
+                conf = 1
+            if conf < model_configs['ols_thres']:
+                continue
+            rng = rng_grid[ridx]
+            agl = agl_grid[aidx]
+            if rng > test_configs['rr_max'] or rng < test_configs['rr_min']:
+                continue
+            if agl > math.radians(test_configs['ra_max']) or agl < math.radians(test_configs['ra_min']):
+                continue
+            obj_dict = dict(
+                id=id + 1,
+                frame_id=frameid,
+                range=rng,
+                angle=agl,
+                range_id=ridx,
+                angle_id=aidx,
+                class_id=classid,
+                score=conf
+            )
+            dts[frameid, classid].append(obj_dict)
+
+    return dts, n_frame
+
+
+def load_vgg_res(filename, config_dict):
+    n_class = config_dict['class_cfg']['n_class']
+    classes = config_dict['class_cfg']['classes']
+    model_configs = config_dict['model_cfg']
+    rng_grid = config_dict['mappings']['range_grid']
+    agl_grid = config_dict['mappings']['angle_grid']
+    test_configs = config_dict['test_cfg']
+
+    with open(filename, 'r') as df:
+        data = df.readlines()
+
+    n_frame = int(float(data[-1].rstrip().split()[0])) + 1
+    dts = {(i, j): [] for i in range(n_frame) for j in range(n_class)}
+
+    for id, line in enumerate(data):
+        if line is not None:
+            line = line.rstrip().split()
+            frameid, ridx, aidx, classid, conf = line
+            frameid = int(frameid)
+            classid = int(classid)
+            ridx = int(ridx)
+            aidx = int(aidx)
+            conf = float(conf)
+            if conf > 1:
+                conf = 1
+            # if conf < 0.5:
+            #     continue
+            rng = rng_grid[ridx]
+            agl = agl_grid[aidx]
+            if rng > test_configs['rr_max'] or rng < test_configs['rr_min']:
+                continue
+            if agl > math.radians(test_configs['ra_max']) or agl < math.radians(test_configs['ra_min']):
+                continue
+            obj_dict = {'id': id + 1, 'frameid': frameid, 'range': rng, 'angle': agl, 'ridx': ridx, 'aidx': aidx,
+                        'classid': classid, 'score': conf}
+            dts[frameid, classid].append(obj_dict)
+
+    return dts, n_frame